2 MICROECONOMICS Competition, Conflict, and Coordination S A M U E L B O W L E S & S I M O N D. H A L L I D AY S A M U E L B O W L E S & S I M O N D. H A L L I D AY MICROECONOMICS: COMPETITION, C O N F L I C T, & C O O R D I N AT I O N 2 This text is provisional and under active revision. Please do not quote without permission. Samuel Bowles (PhD, Economics, Harvard University) heads the Behavioral Sciences Program at the Santa Fe Institute. He has taught microeconomic theory to undergraduates and PhD candidates at Harvard University, the University of Massachusetts, and the University of Siena. He is the author or co-author of Notes and Problems in Microeconomic Theory (1980), Microeconomics: Behavior, Institutions and Evolution (2005), and with the global CORE team, The Economy (2017) and Economy, Society and Public Policy (2019), both open-access introductions to economics (for majors and nonmajors respectively). His research has appeared in the American Economic Review, Nature, Science, Journal of Political Economy, Quarterly Journal of Economics, and Econometrica. Simon Halliday (PhD, Economics, University of Siena, Italy) is an Associate Professor in the Economics department at the University of Bristol. He has Figure 1: Sam at the Santa Fe Institute also taught graduate and undergraduate students at Smith College, the University of Cape Town and Royal Holloway, University of London. His research in experimental economics, behavioral economics and economics education has been published in the Journal of Economic Behavior and Organization, Journal of Behavioral and Experimental Economics, the Journal of Economic Education, and elsewhere. Figure 2: Simon at Smith College Contents Preface 11 I People, Economy and Society 17 1 Society: Coordination Problems & Economic Institutions 2 23 1.1 Societal coordination: The classical institutional challenge . . . . . . . . . . . . . . . . . . . . . .24 1.2 The institutional challenge today . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27 1.3 Anatomy of a coordination problem: The tragedy of the commons . . . . . . . . . . . . . . . . . . .28 1.4 Institutions: Games and the rules of the game . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30 1.5 Over-exploiting nature: Illustrating the basics of game theory . . . . . . . . . . . . . . . . . . . . .33 1.6 Predicting economic outcomes: The Nash equilibrium 1.7 Evaluating outcomes: Pareto-comparisons and Pareto-efficiency . . . . . . . . . . . . . . . . . . .41 1.8 Strengths and shortcomings of Pareto efficiency as an evaluation of outcomes . . . . . . . . . . . .43 1.9 Conflict and common interest in a Prisoners’ Dilemma . . . . . . . . . . . . . . . . . . . . . . . . .44 1.10 Coordination successes: An invisible hand game . . . . . . . . . . . . . . . . . . . . . . . . . . .49 1.11 Assurance Games: Win-win and lose-lose equilibria . . . . . . . . . . . . . . . . . . . . . . . . . .50 1.12 Disagreement Games: Conflict about how to coordinate . . . . . . . . . . . . . . . . . . . . . . . .53 1.13 Why history (sometimes) matters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .54 1.14 Application: Segregation as a Nash Equilibrium among people who prefer integration . . . . . . . .56 1.15 How institutions can address coordination problems . . . . . . . . . . . . . . . . . . . . . . . . . .61 1.16 Game theory and Nash equilibrium: Importance and caveats . . . . . . . . . . . . . . . . . . . . .63 1.17 Application: Cooperation and conflict in practice . . . . . . . . . . . . . . . . . . . . . . . . . . . .65 1.18 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .67 . . . . . . . . . . . . . . . . . . . . . . . .36 People: Self-interest and Social Preferences 2.1 Preferences, beliefs and constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .72 2.2 Taking risks: Payoffs and probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78 2.3 Expected payoffs and the persistence of poverty . . . . . . . . . . . . . . . . . . . . . . . . . . . .81 2.4 Decision-making under uncertainty: Risk-dominance . . . . . . . . . . . . . . . . . . . . . . . . .84 2.5 Sequential games: When order of play matters . . . . . . . . . . . . . . . . . . . . . . . . . . . .87 2.6 First-mover advantage in a sequential game . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89 2.7 Social preferences: Blame Economic man? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .91 2.8 Experiments on economic behavior 2.9 The Ultimatum Game: Reciprocity and retribution . . . . . . . . . . . . . . . . . . . . . . . . . . .95 2.10 A global view: Common patterns and cultural differences . . . . . . . . . . . . . . . . . . . . . . .98 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93 71 4 3 4 5 2.11 The Public Goods Game . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 . 2.12 Application: Evidence from Public Goods Games . . . . . . . . . . . . . . . . . . . . . . . . . . 104 . 2.13 Social preferences are not "Irrational" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 . 2.14 Application. The lab and the street . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 . 2.15 Application: A fine is a price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 . 2.16 Complexity: diverse, versatile, and changeable people 2.17 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 . . . . . . . . . . . . . . . . . . . . . . . . 110 . Doing the best you can: Constrained optimization 3.1 Time: A scarce resource . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 . 3.2 Utility functions and preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 . 3.3 Indifference curves: Graphing preferences 3.4 Marginal utility and the marginal rate of substitution . . . . . . . . . . . . . . . . . . . . . . . . . 126 . 3.5 Application: Homo economicus with Cobb-Douglas utility . . . . . . . . . . . . . . . . . . . . . . 132 . 3.6 The feasible set of actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 . 3.7 The marginal rate of transformation and opportunity cost . . . . . . . . . . . . . . . . . . . . . . 137 . 3.8 Constrained utility maximization: The mrs = mrt rule . . . . . . . . . . . . . . . . . . . . . . . . 140 . 3.9 The price-offer curve, willingness to pay, and demand . . . . . . . . . . . . . . . . . . . . . . . . 145 . 3.10 Social preferences and utility maximization 3.11 Application: Environmental trade-offs 3.12 Application: Optimal abatement of environmental damages . . . . . . . . . . . . . . . . . . . . . 154 . 3.13 Cardinal inter-personally comparable utility: Evaluating policies to reduce inequality . . . . . . . . 159 . 3.14 Application: Cardinal utility and subjective well-being . . . . . . . . . . . . . . . . . . . . . . . . 162 . 3.15 Preferences, beliefs, and constraints: An assessment . . . . . . . . . . . . . . . . . . . . . . . . 164 . 3.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 . 117 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 . Property, Power, & Exchange: Mutual Gains & Conflicts 4.1 Mutual gains from trade: Conflict and coordination . . . . . . . . . . . . . . . . . . . . . . . . . . 172 . 4.2 Feasible allocations: The Edgeworth box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 . 4.3 The Pareto-efficient set of feasible allocations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 . 4.4 Adam Smith’s Impartial Spectator suggests a fair outcome . . . . . . . . . . . . . . . . . . . . . 182 . 4.5 Property rights and participation constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 . 4.6 Symmetrical exchange: Trading into the Pareto-improving lens . . . . . . . . . . . . . . . . . . . 190 . 4.7 Bargaining power: Take-it-or-leave-it . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 . 4.8 Application: Bargaining over wages and hours . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 . 4.9 Application. The rules of the game determine hours and wages . . . . . . . . . . . . . . . . . . . 200 . 4.10 First-mover advantage: Price-setting power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 . 4.11 Setting the price subject to an incentive compatibility constraint . . . . . . . . . . . . . . . . . . . 209 . 4.12 Application. Other-regarding preferences: Allocations among friends . . . . . . . . . . . . . . . . 212 . 4.13 The rules of the game and the problem of limited information . . . . . . . . . . . . . . . . . . . . 217 . 4.14 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 . Coordination Failures & Institutional Responses 5.1 Common property resources, public goods, and club goods . . . . . . . . . . . . . . . . . . . . . 225 . 5.2 A common property resources problem: Preferences . . . . . . . . . . . . . . . . . . . . . . . . 228 . 171 223 5 5.3 Technology and environmental limits: The source of a coordination failure . . . . . . . . . . . . . 232 . 5.4 A best response: Another constrained optimization problem . . . . . . . . . . . . . . . . . . . . . 235 . 5.5 A best-response function: Interdependence recognized . . . . . . . . . . . . . . . . . . . . . . . 238 . 5.6 How will the game be played? A symmetric Nash equilibrium . . . . . . . . . . . . . . . . . . . . 241 . 5.7 How would the players get to the Nash equilibrium? A dynamic analysis . . . . . . . . . . . . . . 244 . 5.8 Evaluating outcomes: Participation constraints, Pareto improvements and Pareto-efficiency . . . . 247 . 5.9 A Pareto inefficient Nash equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 . 5.10 A benchmark socially-optimal allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 . 5.11 Government policies: Regulation and taxation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 . 5.12 Private ownership: Permits and employment 5.13 Community: Repeated interactions and altruism . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 . 5.14 Application: Is inequality a problem or a solution? . . . . . . . . . . . . . . . . . . . . . . . . . . 274 . 5.15 Over-exploitation of a non-excludable resource . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 . 5.16 The rules of the game matter: Alternatives to over-exploitation . . . . . . . . . . . . . . . . . . . 282 . 5.17 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 . II Markets for Goods and Services 291 6 Production: Technology and Specialization 295 7 6.1 The division of labor, specialization and the market . . . . . . . . . . . . . . . . . . . . . . . . . 296 . 6.2 Production functions with a single input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 . 6.3 Economies of scale and the feasible production set . . . . . . . . . . . . . . . . . . . . . . . . . 300 . 6.4 Economies of scale, specialization and exchange . . . . . . . . . . . . . . . . . . . . . . . . . . 303 . 6.5 Comparative and absolute advantage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 . 6.6 Specialization according to comparative advantage . . . . . . . . . . . . . . . . . . . . . . . . . 311 . 6.7 History, specialization, and coordination failures . . . . . . . . . . . . . . . . . . . . . . . . . . . 314 . 6.8 Application: The limits of specialization and comparative advantage 6.9 Production technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319 . 6.10 Production functions with more than one input . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322 . 6.11 Cost-minimizing technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327 . 6.12 Technical change and innovation rents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332 . 6.13 Application: What does the model of innovation miss? . . . . . . . . . . . . . . . . . . . . . . . . 334 . 6.14 Characterizing technologies and technical change . . . . . . . . . . . . . . . . . . . . . . . . . . 336 . 6.15 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339 . . . . . . . . . . . . . . . . . 317 . Demand: Willingness to pay and prices 343 7.1 The budget set, indifference curves and the rules of the game. . . . . . . . . . . . . . . . . . . . 346 . 7.2 Income, prices and offer curves 7.3 Cobb-Douglas utility and demand 7.4 Application. Doing the best you can dividing your time . . . . . . . . . . . . . . . . . . . . . . . . 357 . 7.5 Application: Social comparisons, work hours and consumption as a social activity . . . . . . . . . 360 . 7.6 Quasi-linear utility and demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365 . 7.7 Price changes: income and substitution effects . . . . . . . . . . . . . . . . . . . . . . . . . . . 371 . 7.8 Application: Income and substitution effects of a carbon tax and citizen dividend . . . . . . . . . . 374 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 . 6 8 9 7.9 Application: Giffen Goods and The Law of Demand . . . . . . . . . . . . . . . . . . . . . . . . . 377 . 7.10 Market demand and price elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379 . 7.11 Application. Empirical estimates of the effect of price on demand. . . . . . . . . . . . . . . . . . . 384 . 7.12 Consumer surplus and interpersonal comparisons of utility . . . . . . . . . . . . . . . . . . . . . 386 . 7.13 Application: The effect of a sugar tax on consumer surplus . . . . . . . . . . . . . . . . . . . . . 389 . 7.14 Application. Willingness to pay (for an integrated neighborhood) 7.15 Application: Market dynamics and segregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397 . 7.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401 . . . . . . . . . . . . . . . . . . . 393 . Supply: Firms’ costs, output and profit 405 8.1 Costs of production: An owner’s eye view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408 . 8.2 Accounting profits and economic profits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409 . 8.3 Cost functions: Decreasing and increasing average costs . . . . . . . . . . . . . . . . . . . . . . 411 . 8.4 Application: Evidence about cost functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413 . 8.5 A monopolistic competitor selects an output level . . . . . . . . . . . . . . . . . . . . . . . . . . 417 . 8.6 Profit maximization: marginal revenues and marginal costs . . . . . . . . . . . . . . . . . . . . . 423 . 8.7 The markup, the price elasticity of demand, and entry barriers 8.8 Application: Evidence on the markup in drug prices . . . . . . . . . . . . . . . . . . . . . . . . . 433 . 8.9 Willingness to sell: Capacity constraints and market supply . . . . . . . . . . . . . . . . . . . . . 435 . 8.10 Economic profits and the market supply curve 8.11 Perfect competition among price-taking buyers and sellers: Shared gains from exchange . . . . . 440 . 8.12 The effects of a tax: Consumer surplus, profits, tax revenues and deadweight loss 8.13 Competition among price takers: An assessment . . . . . . . . . . . . . . . . . . . . . . . . . . 446 . 8.14 Two benchmark models of the profit-maximizing firm: Price takers and price makers. 8.15 Application: Dynamics – The growth of firms and the survival of competition . . . . . . . . . . . . 450 . 8.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453 . . . . . . . . . . . . . . . . . . . . 429 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438 . . . . . . . . . 444 . . . . . . . . 448 . Competition, Rent-seeking & Market Equilibration 457 9.1 Modelling the continuum of competition: From one firm to many . . . . . . . . . . . . . . . . . . . 458 . 9.2 Reviewing the monopoly case, n = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463 . 9.3 Duopoly: Two firms’ best responses and the Nash equilibrium 9.4 Oligopoly and "unlimited competition": From a few firms to many firms . . . . . . . . . . . . . . . 472 . 9.5 The extent of competition and the markup over costs . . . . . . . . . . . . . . . . . . . . . . . . 476 . 9.6 Barriers to entry and the equilibrium number of firms . . . . . . . . . . . . . . . . . . . . . . . . 477 . 9.7 A conflict of interest: Profits, consumer surplus, and the degree of competition . . . . . . . . . . . 482 . 9.8 Limited competition and inefficiency: Deadweight loss 9.9 Coordination among firms: Duopoly and cartels . . . . . . . . . . . . . . . . . . . . . . . . . . . 487 . 9.10 Perfect price discrimination: Eliminating deadweight loss at a cost to consumers 9.11 Application: Price discrimination in action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494 . 9.12 Rent-seeking, price-making, and market equilibration . . . . . . . . . . . . . . . . . . . . . . . . 496 . 9.13 Application: When rent-seeking does not equilibrate a market – A housing bubble . . . . . . . . . 501 . 9.14 How competition works: The forces of supply and demand . . . . . . . . . . . . . . . . . . . . . 503 . 9.15 The "perfect competitor:" Rent-seeking firms competing in and for markets 9.16 Application: Declining competition and public policy . . . . . . . . . . . . . . . . . . . . . . . . . 510 . . . . . . . . . . . . . . . . . . . . 463 . . . . . . . . . . . . . . . . . . . . . . . . 483 . . . . . . . . . . 490 . . . . . . . . . . . . . 506 . 7 9.17 III Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512 . Markets with Incomplete Contracting 517 10 Information: Contracts, Norms & Power 523 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523 . 10.2 Incomplete contracts: "... not everything is in the contract" 10.3 Principals and agents: Hidden actions and hidden attributes 10.4 Hidden attributes and adverse selection: The Lemons Problem . . . . . . . . . . . . . . . . . . . 530 . 10.5 Application: Health insurance 10.6 Hidden actions and moral hazards: A contingent renewal contract 10.7 The value of the transaction to the agent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539 . 10.8 The agent’s best response: An incentive compatibility constraint . . . . . . . . . . . . . . . . . . 545 . 10.9 The principal’s cost minimization and the Nash equilibrium . . . . . . . . . . . . . . . . . . . . . 549 . 10.10 Short-side power in principal-agent relationships 10.11 A comparison with complete contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558 . 10.12 Features of equilibria with incomplete contracts: Summing up . . . . . . . . . . . . . . . . . . . . 562 . 10.13 Incomplete contracts and the distribution of gains from exchange 10.14 Application: Complete contracts in the gig economy . . . . . . . . . . . . . . . . . . . . . . . . . 568 . 10.15 Application: Norms in markets with incomplete contracts . . . . . . . . . . . . . . . . . . . . . . 570 . 10.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573 . . . . . . . . . . . . . . . . . . . . . . 524 . . . . . . . . . . . . . . . . . . . . . 527 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534 . . . . . . . . . . . . . . . . . . 536 . . . . . . . . . . . . . . . . . . . . . . . . . . . 555 . . . . . . . . . . . . . . . . . . 564 . 11 Work, Wages & Unemployment 577 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577 . 11.2 Employment as a principal-agent relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578 . 11.3 Nash equilibrium wages, effort, and hiring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581 . 11.4 The employer’s profit-maximizing level of hiring . . . . . . . . . . . . . . . . . . . . . . . . . . . 584 . 11.5 Comparing the incomplete and complete contracts cases . . . . . . . . . . . . . . . . . . . . . . 590 . 11.6 Employment rents and the workers’ fallback option 11.7 Connecting micro to macroeconomics: A no-shirking condition . . . . . . . . . . . . . . . . . . . 599 . 11.8 Incomplete contracts & the distribution of gains from exchange . . . . . . . . . . . . . . . . . . . 603 . 11.9 Application: Contract enforcement technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 604 . 11.10 Equilibrium unemployment and the wage curve . . . . . . . . . . . . . . . . . . . . . . . . . . . 607 . 11.11 The whole-economy model: Profits, wages, and employment . . . . . . . . . . . . . . . . . . . . 611 . 11.12 Monopsony, the cost of inputs and the level of hiring . . . . . . . . . . . . . . . . . . . . . . . . . 595 . . . . . . . . . . . . . . . . . . . . . . . . . 617 . 11.13 Monopsony and the cost of hiring (non-shirking) labor 11.14 The effects of a minimum wage on hiring and labor earnings . . . . . . . . . . . . . . . . . . . . 624 . . . . . . . . . . . . . . . . . . . . . . . . 621 . 11.15 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630 . 12 Interest, Credit & Wealth Constraints 635 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635 . 12.2 Evidence on credit and wealth constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637 . 12.3 The wealthy owner-operator case 12.4 Complete credit contracts: A limiting case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641 . 8 IV 12.5 The general case: incomplete credit contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650 . 12.6 The Nash equilibrium level of risk and interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654 . 12.7 Characteristics of the incomplete contract Nash equilibrium 12.8 Many lenders: Competition and barriers to entry 12.9 Wealth matters: Borrowing with equity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 667 . 12.10 Excluded and credit-constrained borrowers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670 . 12.11 Why redistributing wealth may enhance efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . 672 . 12.12 Competition, barriers to entry and the distribution of rents . . . . . . . . . . . . . . . . . . . . . . 676 . 12.13 Application: From micro to macro: The multiplier and monetary policy 12.14 Application. Why cotton became king in the U.S. South following the end of slavery . . . . . . . . 684 . 12.15 Why and How Wealth Matters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685 . 12.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687 . . . . . . . . . . . . . . . . . . . . . 659 . . . . . . . . . . . . . . . . . . . . . . . . . . . 663 . . . . . . . . . . . . . . . . 678 . Economic systems and policy 691 13 A Risky & Unequal World 695 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695 . 13.2 Choosing Risk: Gender differences 13.3 Risk preferences over lotteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 699 . 13.4 Wealth differences and decreasing risk aversion . . . . . . . . . . . . . . . . . . . . . . . . . . . 703 . 13.5 Application: Risk, wealth and the choice of technology 13.6 Doing the best you can in a risky world . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 708 . 13.7 How risk aversion can perpetuate economic inequality . . . . . . . . . . . . . . . . . . . . . . . 712 . 13.8 How insurance can mitigate risk and reduce inequality . . . . . . . . . . . . . . . . . . . . . . . 714 . 13.9 Buying and selling risk: Two sides of an insurance market . . . . . . . . . . . . . . . . . . . . . . 720 . 13.10 Application: Free tuition with an income-contingent tax on graduates . . . . . . . . . . . . . . . . 725 . 13.11 Another form of insurance: A linear tax and lump sum transfer . . . . . . . . . . . . . . . . . . . 730 . 13.12 A citizen’s preferred level of tax and transfers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735 . 13.13 Political rents: Conflicts of interest over taxes and transfers . . . . . . . . . . . . . . . . . . . . . 739 . 13.14 Application: Choosing justice, a question of ethics 13.15 Risk, uncertainty and loss aversion: Evaluation of the model . . . . . . . . . . . . . . . . . . . . 745 . 13.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697 . . . . . . . . . . . . . . . . . . . . . . . . 706 . . . . . . . . . . . . . . . . . . . . . . . . . . 740 . 14 Perfect Competition & the Invisible Hand 751 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751 . 14.2 A general competitive equilibrium 14.3 Market clearing and Pareto-efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757 . 14.4 Prices as messages, markets as information processors 14.5 The Fundamental Theorems and Pareto efficiency 14.6 Perfectly competition and inequality: Distributional neutrality 14.7 Market failures due to uncompensated external effects . . . . . . . . . . . . . . . . . . . . . . . 772 . 14.8 Market dynamics: Getting to an equilibrium and staying there . . . . . . . . . . . . . . . . . . . . 775 . 14.9 Bargaining and rent-seeking: A more realistic model of market dynamics . . . . . . . . . . . . . . 778 . 14.10 Disequilibrium trading creates inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 781 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753 . . . . . . . . . . . . . . . . . . . . . . . 761 . . . . . . . . . . . . . . . . . . . . . . . . . . 764 . . . . . . . . . . . . . . . . . . . . . 766 . 9 14.11 Bargaining to an efficient outcome: The Coase Theorem . . . . . . . . . . . . . . . . . . . . . . 784 . 14.12 An example: How Coasean bargaining works . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787 . 14.13 Application: Bargaining over a curfew . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794 . 14.14 Bargaining, markets, and public policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803 . 14.15 Application: Planning vs the market in the history of economics 14.16 Perfect competition or the perfect competitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807 . 14.17 Conclusion: Ideal systems in an imperfect world . . . . . . . . . . . . . . . . . . . . . . . . . . . 808 . . . . . . . . . . . . . . . . . . . 805 . 15 Capitalism: Innovation & Inequality 813 15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 813 . 15.2 Capitalism’s success: The hockey stick of history . . . . . . . . . . . . . . . . . . . . . . . . . . 815 . 15.3 Capitalism and inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 817 . 15.4 Employment as insurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 818 . 15.5 Explaining the hockey stick: Capitalist firms share risks & promote innovation 15.6 How can more equal societies also be innovative? . . . . . . . . . . . . . . . . . . . . . . . . . . 824 . 15.7 Measuring economic inequality: The Gini coefficient and the Lorenz curve . . . . . . . . . . . . . 826 . 15.8 Inequality and the macro-economy: A micro-economic explanation . . . . . . . . . . . . . . . . . 830 . 15.9 Market power and the distribution of income . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 834 . 15.10 Modern monopoly, winners-take-all and public policy 15.11 Application: Public policy to raise wages and reduce unemployment and inequality . . . . . . . . . 839 . 15.12 Application: Trade unions, inequality, and economic performance . . . . . . . . . . . . . . . . . . 844 . 15.13 Capitalism as an economic and social order: Disparities in wealth and power . . . . . . . . . . . . 846 . 15.14 Would a wealth-poor person want to hold a risky asset? . . . . . . . . . . . . . . . . . . . . . . . 852 . 15.15 Risk, redistribution and innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855 . 15.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 856 . . . . . . . . . . . . 820 . . . . . . . . . . . . . . . . . . . . . . . . . 836 . 16 Public policy and mechanism design 16.1 Mechanism design: Policy implementation by Nash equilibrium . . . . . . . . . . . . . . . . . . . 863 . 16.2 Optimal contracts: internalizing external effects of public goods . . . . . . . . . . . . . . . . . . . 866 . 16.3 The social multiplier of cigarette taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 873 . 16.4 The theory of the second best and public policy . . . . . . . . . . . . . . . . . . . . . . . . . . . 879 . 16.5 Deception as an impediment to efficient exchange . . . . . . . . . . . . . . . . . . . . . . . . . . 882 . 16.6 When optimal contracts fail: The case of team production . . . . . . . . . . . . . . . . . . . . . . 887 . 16.7 The limits of incentives: Crowding out and crowding in. . . . . . . . . . . . . . . . . . . . . . . . 895 . 16.8 Beyond market versus government: Expanding the space for policies and institutions . . . . . . . 901 . 16.9 Application: A worker-owned cooperative. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 901 . 16.10 The distributional impact of public policies: Rent control . . . . . . . . . . . . . . . . . . . . . . . 904 . 16.11 Egalitarian redistribution to address market failures . . . . . . . . . . . . . . . . . . . . . . . . . 911 . 16.12 Why governments sometimes fail: A caveat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 912 . 16.13 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913 . 861 Glossary 917 Glossary 919 10 17 End notes 937 Preface To its 18th and early 19th century founders, the subject of economics was the wealth of nations and people. This was no less true of Karl Marx, the most famous critic of capitalism, than it was of Adam Smith’s whose The Wealth of Nations is considered the most powerful defence of the then emerging capitalist economic system. Economics was at the time called political economy, and it sought to understand how and why society was being transformed as a result of capitalism, a novel way of organizing how people produce, exchange and distribute the things we live on. Capitalism continues to change the world, and the task of economics is to understand this process, and how our economies might be made to work better for people today and in the future. Welcome to Microeconomics: Competition, Conflict, and Coordination and best wishes for your journey through its content. Let’s begin by saying how we came to think that economics is important and then explaining our strategy for how you can best learn to do economics. Economics engaged in the world Contrary to its reputation among students for being remote from reality, economics has always been about changing the way the world works. The earliest economists – the Physiocrats in late 18th century France and the Mercantilists before them – were advisers to kings and queens of Europe. Today’s macroeconomic managers, economic development advisors and advocates of competing policies concerning intellectual property rights or the global movements of goods and people continue this tradition of real world engagement. Economists have never been strangers to policy-making, constitution building and attempts at economic reform for the betterment of people’s living conditions. Alfred Marshall’s (1842-1924) Principles of Economics, published in 1890 was the first great text in what came to be called neoclassical economics. It opens with these lines: 12 MICROECONOMICS - DRAFT “Now at last we are setting ourselves seriously to inquire whether .. there need be large numbers of people doomed from their birth to hard work in order to provide for others the requisites of a refined and cultured life, while they themselves are prevented by their poverty and toil from having any share or part in that life. ...[T]he answer depends in a great measure upon facts and inferences, which are within the province of economics; and this is it which gives to economic studies their chief and their highest interest.” The hope that economics might assist in alleviating poverty and securing the conditions under which free people might flourish is at once economics’ most inspiring calling and its greatest challenge. Like many, both of us were drawn to economics by this hope. One of us (Simon) grew up in Cape Town under the system of racial segregation called apartheid. He vividly remembers the demonstrations that finally brought that system down and the long lines of people waiting to vote in South Africa’s first democratic elections in 1994. He volunteered in the poor townships surrounding Cape Town teaching critical thinking and debating, skills required to make the new democracy work. . Having initially followed his passion for theater and poetry, he switched into economics to gain the analytical tools to understand and address his country’s challenges. The other of your authors (Sam), having been a schoolboy in India and a secondary school teacher in Nigeria before turning to economics, naturally came to the field expecting that it would address the enduring problem of global poverty and inequality. At age eleven Sam had noticed how very average he was among his classmates at the Delhi Public School – in sports, in school work, in just about everything. A question that he then asked his mother has haunted him since: "how does it come about that Indians are so much poorer than Americans, given that as people we are so similar in our abilities?" And so he entered graduate school hoping that economics might, for example, explain why workers in the United States produced almost as much in a month as those in India produce in a year, and why the Indian population was correspondingly poor. We now know that the many conventional economic explanations for the gap in standards of living between the two countries are part of the answer but far from all of it: by any reasonable accounting, the difference in the amount of machinery, land and other capital goods per worker and in the level of schooling of the U.S. and Indian work forces explain much less than half of the difference in output per hour of work. It seems likely that much of the unexplained difference results from causes that until recently have been less studied by economists but which are a central theme of this book. Chief among these are differences in institutions, that is differences in how the activities of the millions of actors in the two C O N T E NTS 13 economies are coordinated by some combination of markets, private property, social norms, and governments. What should economics be about? We do not think that we are atypical – either among our economics colleagues, or our students, or for that matter among people generally – in our hope that economics can contribute to improving the way these institutions work. The CORE Team – a group of economic researchers and teachers who have created an open access introductory economics course (www.coreecon.org) – posed the following question to students around the world on the first day of their introductory classes: "what is the most pressing problem economists today should be addressing?" The results from a total of 4,442 students from 25 universities in twelve countries over the years 2016-18 are summarized in the word cloud in Figure 3. The themes are remarkably consistent across universities and countries. Unemployment, inflation, and growth, all important topics in most macroeconomics courses, are on the minds of students. But inequality (along with "poverty") is the overwhelmingly dominant issue. Environmental sustainability (and "climate change"), the future of work (robots, digitalization), globalization and migration, innovation, financial instability; and how governments work ("corruption," "war") are also present. A few students also identified particular political events, like the "Brexit" referendum in 2016, that favored the U.K. leaving the European Union, as problems that economics should address. In word clouds based on more recent surveys "climate change" is as large a concern as "inequality" among students. Figure 3: Student replies to the question "What is the most pressing problem economists should be addressing?" The size of the font is proportional to the frequency with which subjects mentioned the word or term. Surprisingly professional economists at the New Zealand Treasury and central bank and new hires at the Bank of England responded very similarly to students.The less frequently mentioned – smaller font– topics are more readable in the individual word clouds from each of the 25 samples of students that you can access at https://tinyco.re/6235473 The microeconomic theory that you will learn has a lot to say about these issues. Included are tried and true workhorse concepts that you have probably 14 MICROECONOMICS - DRAFT already encountered, like opportunity costs, mutual gains from exchange, constrained optimization and trade offs. Also essential in understanding issues like those in the word cloud are concepts that have more recently risen to prominence among economists. Examples are the importance of cooperation and social (rather than entirely selfish) motivations and modeling strategic interactions among people, including conflicts over the distribution of the mutual gains from exchange. "If you are not doing something, you are not learning anything!" The phrase just above is our motto when it comes to learning. Economics is not just something you learn. It is something you do. Think of studying economics as learning a new language. Mastering a large vocabulary and the grammatical rules is essential, but it is not the same as speaking the language. The test of what you have learned after studying this book is not just what you know, but what you can do with it. Doing economics is what you can say or write – the case you can make for or against a proposed economic policy, the analysis of the reasons for some new development in the global economy – in other words what you can do as a result of what you know. Like mastering a new language, doing economics is essential to learning the subject. And also like a language, you will learn to do economics more readily if you have a clear need to know. We begin each chapter with a real world problem or example that can be better understood using the concepts and models to be introduced in the chapter. These opening paragraphs suggest the need to know what is to follow. The empirical examples also serve as a reminder that the point to the model is to understand the world; and as we proceed through chapters we will ask: how good a job does this particular model do in that respect? In the margin at the beginning of each chapter is a set of learning objectives phrased as new capacities to do things that most likely you were unable to do before. We place great emphasis on your ability to solve problems in which there are right and wrong answers. But it is also important to learn how to formulate arguments and hypotheses about questions that are thus far unanswered, some of which may remain so, and to express economically informed opinions on issues that will continue to be debated due to the fact that people’s values differ. Interspersed with the contents of the chapters, but offset by boxes, are two important resources: Mathematics Notes M-notes contain the details of mathematical derivations C O N T E NTS and other analysis as well as worked examples that illustrate the mathematical models in the text. Checkpoints are self-tests to confirm that you understand the content of the section. The first step in "doing economics" is by checking your understanding of the passage you have just read. At the end of each chapter you will find the following: Important Ideas The main ideas in each chapter are provided in a table. At the end of the book, you will also find that all the definitions of the book are included as a glossary for you to consult and improve your understanding. Mastering the use of these terms is essential to doing economics. Try using each of them in a complete sentence of our own. Making connections provides some guidance in seeing how the ideas in each chapter are connected to each other and to other themes in the book, so that you will be able to draw together the ‘big picture’ about the main messages and themes of the book. Try restating these connections making use of the terms in Important Ideas. Or better yet: make a mind map using the Important Ideas and Making Connections features. Mathematical Notation The book contains a variety of important mathematical tools to help model the various economic ideas in the book. To assist you with your reading of each chapter and to understand better each model you encounter, we provide a table of the mathematical notation you will encounter in that chapter. There is also a complete list of the notation used at the end of the book. We use the margins of the book for a variety of purposes: Definitions We define important terms in the margins where they first are introduced. All of the definitions are collected in a glossary. Reminders We put reminders in the text often to help you to see the connections of ideas throughout the book. Example An example will often illustrate an idea with a relevant example of a person, firm, or country making decisions that are similar to those described in the text. Fact Check When we need to verify or illustrate an idea with data or an empirical example we will do so with a Fact Check. History These introduce you to some of those people who have contributed to economics or to relevant historical facts. M-check If an idea requires a brief mathematical clarification that does not require it’s own M-Note, then we may convey that in a margin note. 15 16 MICROECONOMICS - DRAFT Economics is an integrated body of knowledge, and it is best learned in a cumulative way, mastering a set of concepts and going on to use those concepts in mastering additional concepts. What this means, practically is that it is best to study earlier chapters before moving on to later ones. Sections labeled "application" however provide illustrations of how the ideas and models being taught in a particular chapter can be used, and these do not introduce new material that is essential to the chapters that follow. Micoeconomics is waiting for you. Just do it! Samuel Bowles and Simon Halliday Santa Fe Institute, Santa Fe, New Mexico, U.S.A, and Smith College, Northampton, Massachusetts, U.S.A. Part I People, Economy and Society 19 The man . . . enamored of his own ideal plan of government, . . . seems to imagine that he can arrange the different members of a great society with as much ease as the hand arranges the different pieces upon a chess-board . . . but . . . in the great chess-board of human society, every single piece has a principle of motion of its own, altogether different from that which the legislature might choose to impress upon it. If those two principles coincide and act in the same direction, the game of human society will go on easily and harmoniously, and is very likely to be happy and successful. If they are opposite or different, the game will go on miserably, and the society must be at all times in the highest degree of disorder. Adam Smith,Theory of Moral Sentiments, 1759,Part VI, Section 1 As individuals, our physical capacities are hardly remarkable compared to other animals. But by coordinating with others – finding ways that our individual efforts can add up to a whole that is more than the sum of its parts – humans are unique as a species, engaging in common pursuits on a global scale and for better or worse, transforming nature and inventing previously unimagined devices and ways of life. Economics provides a lens for studying this social aspect of human uniqueness by analysing how people interact with H I S TO RY What makes humans unique among all the animals is our capacity to cooperate in very large numbers and to adjust the ways that we cooperate to changing circumstances. Here (https://tinyurl.com/y3bpy4px) the Israeli historian Yuval Noah Harari explains why this is so. each other and with our natural surroundings in producing and acquiring our livelihoods. We begin (in Chapter 1) by developing a common framework for studying the various types of social interactions using game theory to pose a question older than economics. This is: how can a society’s institutions – its laws, unwritten rules and social norms – harness individuals’ pursuit of their own objectives to generate common benefits and to avoid outcomes that none would have chosen. The challenge is how to combine freedom – individuals’ pursuit of their own objectives – with the common good, improving the livelihoods of all members of the society. This challenge is called the problem of societal coordination: how we can coordinate – that is organize – our actions to yield desirable results for society? The example of societal coordination we use in Chapter 1 to illustrate this challenge is about the other aspect of economics: how we relate to our natural surroundings, illustrated by a problem of over-exploiting an environmental resource. Adam Smith, considered by many to be the founder of economics, understood the challenge well. And he understood that economics – or "political economy" as it was then called –is fundamentally a social science: it is about how people interact. You can see this in his warning above about the disastrous consequences of treating people as if they were simply chess pieces who could be moved around on the chess board of life at the will of a government, more or less like an engineer might design a machine. An adequate response to the challenge of combining freedom and the com- E CONOMICS is the study of how people interact with each other and with our natural surroundings in producing and acquiring our livelihoods. 20 MICROECONOMICS - DRAFT mon good must therefore be based on knowledg of how individuals act depending on the situation they are in, and how changing the situation will change how they act. We therefore (Chapter 2) turn to individuals and their motives– whether self interested or generous, opportunistic or ethical – explaining how people do the best they can in given situations. In this chapter we consider individuals in situations where they act in isolation rather than interacting with other individuals. But people rarely act in isolation: Economics allows us to understand the sometimes surprising or unintended society-wide effects of when we interact with others, whether it be directly with our own employer or indirectly with literally millions of people involved in producing and distributing the goods making up our livelihoods. A basic insight for this understanding is that we are better off by interacting with others. But our interactions also give rise to conflicts. When people engage with others in buying and selling, working and investing there are mutual benefits potentially available to all parties involved. This must be the case if participation in these and other economic activities is voluntary. But unavoidably there are also conflicts over how these mutual gains are divided (Chapter 4). We evaluate the outcomes of economic interactions by two standards: • Efficiency : the extent to which all of the potential gains are realized (which is how economists use the term efficiency) and • Fairness: whether the distribution of the gains and the process that determines who gets what is just. And we study the various ways that exchanges and other economic activities may be carried out and how they may affect both the efficiency and fairness of the outcome. In our interactions with each other and with nature we frequently fail to exploit all of the potential mutual gains. An example is when a person with the capacity and desire to produce goods and services needed by others cannot find a job. Another is over-exploitation of a fishery or some other environmental resource. These are called coordination failures because they result from inadequacies in the ways that our institutions coordinate the ways that we interact. Coordination failures are often due to our conflicts over the distribution of potential mutual gains or to the fact that when we act we do not take account of the effects of our actions on others (Chapter 5). Markets, government policies, well-designed property rights, a concern for one another’s well being, and communities can help address these coordination failures so that no potential mutual gains remain unexploited, and the distribution of gains is regarded as 21 fair. The final chapter of this book will bring together the concepts, models and other ways of thinking that you have learned and apply them to the challenge of improving the way the economy works by both of these standards: efficiency and fairness. 1 Society: Coordination Problems & Economic Institutions DOING ECONOMICS Two neighbors may agree to drain a meadow, which they possess in common; because ’tis easy for them to know each others mind; and each must perceive, that the immediate consequence of his failing in his part, is the abandoning of the whole project. But ’tis very difficult and indeed impossible, that a thousand persons shou’d agree in any such action; it being difficult for them to concert so complicated a design, and still more difficult for them to execute it; while each seeks a pretext to free himself of the trouble and expense, and wou’d lay the whole burden on others. David Hume, A Treatise of Human Nature, Volume II (1967 [1742]: 304) At the turn of the present century, the process of economic development had bypassed almost all of the two hundred or so families that made up the village of Palanpur in the Indian state of Uttar Pradesh. But for the occasional watch, bicycle or irrigation pump, Palanpur appeared to be a timeless backwater, untouched by India’s cutting edge software industry and booming agricultural regions. Less than a third of the adults were literate, and most had endured the loss of a child to malnutrition or to illnesses that had long been forgotten in other parts of the world. A visitor to the village approached a farmer and his three daughters weeding a small plot of land. The conversation turned to the fact that Palanpur farmers This chapter will enable you to:: • Use game theory to analyze how people interact in the economy, each affecting the conditions under which the others decide how to act. • Understand why the outcomes of interactions are often worse for people than they could be and how interactions might be better organized to improve the quality of people’s lives. • Recognize that unsatisfactory outcomes occur when people fail to coordinate with each other and to take account of the effect of their own actions on others. • Explain how problems like environmental damage and global poverty can be the result of failed coordination. • Represent institutions as "the rules of the game." • Identify that economic institutions determine incentives for people’s behavior and can affect how successfully we address coordination problems. • Explain why when people have limited information and conflicts of interest they often fail to implement ’win-win’ outcomes. plant their winter crops several weeks after the date that would maximize the amount of grain they could get at harvest time. The farmers knew that planting earlier would produce larger harvests, but no one, the farmer explained, wants to be the first farmer to plant their seeds, as the seeds on any lone plot would be quickly eaten by birds. Curious, the visitor asked if a large group of farmers, perhaps members of the Figure 1.1: Palanpur farmers threshing and winnowing grain (separating grain from chaff) . Photo courtesy of Nicholas Stern. 24 MICROECONOMICS - DRAFT same extended family, had ever agreed to plant their seeds earlier, perhaps on the same day to minimize the individual losses. “If we knew how to do that,” the farmer said, looking up from his hoe and making eye contact with the visitor for the first time, “we would not be poor.”1 1.1 Societal coordination: The classical institutional challenge For the Palanpur farmers, the decision when to plant is a coordination problem. A coordination problem is a situation in which people could all be better off, or at least some be better of and none be worse off, if they all jointly decided how to act – that is, if they coordinated their actions – than if they act individually. The planting choice is a coordination problem because: • the farmer does better or worse depending on what other farmers do, • all the farmers would do better if they could coordinate their actions by jointly agreeing to all do what would be mutually beneficial namely, planting early, but • it is a problem because the farmers may not be able to coordinate, and • if they do not coordinate, then all of the farmers will do worse than they all could otherwise have done (that is, had they all planted late). To stress the fact that coordination problems often affect an entire population (even though we explain them using two person examples) we sometimes use the expression societal coordination problems. Notice that one farmer cannot dictate the actions of the other farmers, nor can they come to a common agreement about what to do ("if we knew how to do that, we would not be poor") – the inability to come together and coordinate is at the heart of coordination problems. Our example is about farming, but it could just as well have been about the owners of many firms each producing some different product deciding independently whether to invest in new buildings and equipment. Firms will only invest only if they anticipate that there will be sufficient demand for the resulting increase in their outputs. Each firm’s investment – purchasing new machinery and construction materials, and hiring more employees – means greater demand for the other firms’ products, including the products purchased by the newly employed workers constructing the new offices, factories and the like with their earnings. If they all invest, then the other firms’ investments will create sufficient demand to purchase the output from each of the firms’ expanded capacity. But if one firm expands and the rest do not, then that firm is likely to find that it cannot sell all of what it is now capable of producing. C OORDINATION PROBLEM A coordination problem is a situation in which people could all be better off (or at least some be better of and none be worse off) if they jointly decide how to act – that is, if they coordinate their actions – than if they act independently. For example, deciding on which side of the road to drive is a coordination problem. H I S TO RY In his address accepting the Nobel Prize for economics in 1979, University of Chicago economist T.W. Schultz said: "Most of the people in the world are poor, so if we knew the economics of being poor, we would know much of the economics that really matters." He was right then and he is right now.2 S O C I E T Y : C O O R D I N AT I O N P R O B L E M S The best choice for each firm depends on the choices made by other firms. If they could coordinate – decide jointly that they would all invest – they would all profit, but coming to such an agreement may not be possible. The owners of firms therefore face a coordination problem similar to the problem faced by the Palanpur farmers. Their individual success depends on their ability to coordinate their actions with others. This is not a new problem: it was on a central concern of the founders of economics. David Hume (the 18th century British philosopher and economist quoted at the start of this chapter) used an example – two landowners considering draining a meadow – to pose what he considered the most important problem facing society, namely, devising institutions that would reconcile the pursuit of individual objectives (avoiding the "trouble and expense" in his example of the meadow) with getting desired societal outcomes (improving the value of the meadow by draining it). His simple two-person example was meant to illustrate the need (in a society of "a thousand persons") for a government to address the broader societal coordination problems of his day. Though the term was invented only two centuries after Hume, he was using what we now call game theory to make his case. Let’s apply his reasoning to the farmers of Palanpur. Like Hume we will consider just two farmers as a way of representing the institutional challenge faced by the entire village. Figure 1.2 shows the outcomes for two players, Aram and Bina, choosing when to plant their grain. The figure illustrates the values of the farmers’ crops, which, in a poor village like Palanpur, is the main incentive for farmers, whether they consume the crop themselves or sell it for money income to spend on other things. Each farmer can either plant early or plant late and while (also as in Hume’s example) two people could probably come to some agreement about what to do, remember that we are using this two-person example to illustrate the entire village of about 200 families of farmers, so we assume that they cannot coordinate on some agreed upon actions for the two jointly. There are four possible outcomes: • If both players plant early, they each achieve their best possible harvest, because they grow the most grain through sharing the risk of having their seeds eaten by birds (outcome c in Figure 1.2). • If Aram plants early while Bina plants late, Aram has his seeds eaten by birds and gets no harvest (the worst outcome for him), whereas the late planter gets a good (but not the best) harvest. While none of her seeds are eaten by the birds, planting late is not the best for growing the most grain (outcomes b and d in Figure 1.2). The same is true if Bina planted late when Aram planted early. • If both plant late, they harvest a smaller crop while also sharing the risk of their seeds being eaten, a bad outcome (outcome a in Figure 1.2). & ECONOMIC INSTITUTIONS 25 26 MICROECONOMICS - DRAFT Figure 1.2: Planting in Palanpur. This figure shows "what-if" outcomes for planting in Palanpur. Each column represents a possible combination of Aram planting early or late and Bina planting early or late with the corresponding outcomes being worst, bad, good, or best in terms of how much grain they grow. The people of Palanpur are stuck in the bad outcome even though they would all be better off if they all planted early (they would both move from a "bad" outcome to the "best" outcome in the figure). They are experiencing a co- C OORDINATION FAILURE A coordination failure occurs when people facing a coordination problem fail to coordinate their actions in a way to implement outcome that allows them all to be better off (or at least some to be better off and none to be worse off). ordination failure, namely a coordination problem that is not addressed by appropriate institutions. A modern day David Hume would point out that a government could simply impose a sufficient tax on those planting late to ensure that most farmers would plant early. Adam Smith, a generation after Hume, would stress the value of the exchange of privately owned goods on competitive markets as a way of coordinating the actions of large numbers of people, who would be guided (even without knowing it) by what he termed "an invisible hand." Hume, Smith and the other founders of European political philosophy and political economy posed what we call the classical institutional challenge. These philosophers and economists wanted to know how to design institutions – rules and practices governing peoples’ behavior – so that people could be left free to make their own decisions, and at the same time avoid outcomes that were inferior for everyone. More precisely, how do we design institutions which encourage coordination by free choice while avoiding poor outcomes such as planting late in Palanpur? The 18th and 19th century political economists and philosophers who founded the field of economics were H I S TO RY Adam Smith wrote the following: “[E]very individual [. . .], indeed, neither intends to promote the public interest, nor knows how much he is promoting it [. . .] he intends only his own security; . . . he intends only his own gain, and he is in this . . . led by an invisible hand to promote an end which was no part of his intention . . . By pursuing his own interest he frequently promotes that of the society more effectually than when he really intends to promote it.”3 S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS attempting to provide solutions to coordination problems. Checkpoint 1.1: Planting in Palanpur: A Coordination Problem Imagine that you are Bina in the figure above, and that you did not know whether Aram would plant early or late. What would you do? Suppose you and Aram were neighbors and you could talk with him; what would you say? 1.2 The institutional challenge today The classical institutional challenge remains with us, although some of the forms that it takes today – including global climate change and the appropriate intellectual property rights for sharing digitized knowledge – were unknown to the great 18th and 19th century thinkers. Consider the following coordination problems: • How do we sustain the global environment? To avoid damaging climate change we need to coordinate our reduction of emissions. Many people and firms would prefer that someone else reduce their carbon footprint. How can we address climate change in a way that is both fair and imposes the least possible costs? • How do we make the best use of our ability to create and use knowledge? If we all agree to share the knowledge we have with others we may all be better off: when I transfer my knowledge to you I do not lose the ability to continue using it. But each of us may profit by restricting others’ use of our knowledge by means of patents, copyrights and other intellectual property rights. • How do we move around a city without overcrowding streets and causing delays? My decision whether to drive, walk, or take public transport affects not only my own travel time, but also the degree of traffic congestion and delays experienced by everyone else. Everyone might be better off if the use of private vehicles was substantially reduced, but few will reduce their driving unless other people reduce theirs as well. These are all coordination problems because an outcome that is better for all is possible if people find a way to jointly agree to a course of action. But for reasons we will explain in detail, people also routinely fail to coordinate and suffer bad consequences as a result, including the following: • over use of some resources illustrated by pollution, over-grazing, traffic congestion, and climate change; and • under use of other resources such as the productive capacities and creativity of people and the knowledge that we have created, illustrated by Figure 1.3: Traffic headed out of a major city. Image Credit: Photo by Preillumination SeTh (@7seth). 27 28 MICROECONOMICS - DRAFT unemployment and the enduring poverty of the people of Palanpur and villages like it around the world. Checkpoint 1.2: Coordination Problems You Have Known Think of a social interaction in which you have been involved that was a coordination problem and using the description of why planting in Palanpur is a coordination problem (the bulleted points above) explain why it was a problem and how coordination might have (or did) address the problem. 1.3 Anatomy of a coordination problem: The tragedy of the commons The over use of environmental resources provides a good illustration of why coordination problems arise. In 1968, Garrett Hardin, an ecologist, famously described what he called the tragedy of the commons, an example of a coordination failure.4 He told a story about a group of herders who share a pasture. The pasture was common land – hence a “commons” – shared by many herders. But why was his story a tragedy? Each herder could put as many animals in the pasture as they wished, and overgrazing will lead to erosion and the ruin of the pasture. Hardin reasoned that if the land is common to all and no one herder owns it, each herder has no interest in limiting how many animals they put in the common pasture. A ruined pasture is of no value to any of the herders. But each herder’s selfinterest leads them to neglect the effect their actions have on others. The outcome is a tragedy. With the term tragedy of the commons, Hardin gave social science one of the most evocative metaphors since Adam Smith’s “invisible hand.” Indeed Hardin called his tragedy a “rebuttal to the invisible hand.” The two metaphors are powerful because they capture two essential yet contrasting social insights. When guided by an invisible hand, social interactions reconcile individual choice and socially desirable outcomes. By contrast, the actors in the tragedy of the commons pursue their private objectives to tragic consequences for themselves and others. The natural setting for Hardin’s tragedy was chosen for its imagery. The underlying problem applies to many situations where people typically cannot or do not take account of the effects of their actions on the well-being of others. You can think of a city’s streets as a commons, and people deciding to drive rather than walk, bike, or use public transport as similar to the herders putting cattle on the common. The modern day "tragedy of the roadways" is a traffic jam. T RAGEDY OF THE C OMMONS The tragedy of the commons is a term used to describe a coordination failure arising when a shared resource available for all to use (’the commons’) is over-used so that all users are worse off than they would have been if they had coordinated their actions so that use was restricted. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 29 What are the common elements in Hume’s drain the meadow problem, the farmers in Palanpur planting late, Hardin’s herders overgrazing their pasture and our modern city dwellers clogging the streets with their vehicles? In each of these three cases, the reason why uncoordinated activities of people pursuing their own ends produce outcomes that are worse for all is that each participant’s actions affect the well-being of others but these effects are not taken into account by the individual actors when they decide how to act. These impacts of our actions on others that we do not take account of in deciding what to do are termed external effects. Here are the external effects that actors in our four examples do not take into E XTERNAL E FFECT An external effect occurs when a participant’s action confers a benefit or imposes a cost on other participants and this cost or benefit is not taken into account by the individual taking the action. External effects are also called simply externalities. External effects that result in costs to others are called negative external effects or external diseconomies. External effects that confer benefits on others are called positive external effects, external benefits, or external economies. account when deciding what to do: • The person who lives in a city who drives to work, adds congestion to the streets, and therefore increases the travel time of others. • Hume’s farmer who does not drain the swamp and imposes the cost of doing so on the other farmer. • The Palanpur farmer who plants late, imposes a cost on the other farmer who will have his seeds devoured by birds if he plants early. Likewise the farmer who plants early confers a benefit on the other farmer who can benefit by planting at the right time (early) without severe losses of seed to the birds. • The herder who places additional cattle on the common pasture reduces the grass available to the other herders stock. Addressing coordination problems by internalizing external effects Simply abolishing these and other external effects that are the root of coordination problems is not an option. There is no way to organize society so that nothing that we do would affect others, each person on his or her self sufficient island. Apart from not being much fun, life would be impossible in a society of total social isolates (just think about how the next generation would be born and raised!). So, to address the classical institutional challenge as to prevent or at least minimize coordination failures we need to find ways of inducing each participant to take adequate account of the effects of their actions on others. This is called internalizing an external effect. We use the term external effect because the effect is outside of the individual’s process of decision-making when taking the action. To internalize the external effect, you ensure that the person who acts bears the costs of their negative effects on others and reaps the rewards of their positive effects on others. In this way the otherwise I NTERNALIZATION OF EXTERNAL EFFECTS in economics refers to any way that people can be brought to take appropriate account of the effects of their actions on others. In psychology the term internalization means to to adopt societies values or standards as one’s own values. 30 MICROECONOMICS - DRAFT "external" costs and benefits become part of the individual’s decision-making process, leading them to "take adequate account of the effects of her actions on others." If the “others” are our family, our neighbors, or our friends, our concern for their well-being or our desire to be well regarded by others might get us to take account of the effects of our actions on them. Reflecting this fact, an important response to the classical institutional challenge – one that long predates the classical economists – is that caring for the well-being of others need not be confined to friends and relatives but may extend to all of those with whom we interact. Ethical guides such as the “golden rule” are ways that H I S TO RY The “golden rule” is “to do unto others as you would have them do unto you" (Matthew, 7: 12). Or, treat others as you would like to be treated yourself. The same ethical principle is found in Islamic scriptures and in the teaching of other religions. people often internalize the effects of our actions on others, even when the others are total strangers to us. But, over the past five centuries, people have come to interact not with a few dozen people as humans have for most of our history and pre-history but directly with hundreds and indirectly with millions of strangers. The classical economists in the 18th century were responding to the fact that the generosity or ethical motivations that one might feel towards ones family or neighbors would not be sufficient to induce people to take account of the effect of their actions on others once these external effects spread across the entire network of global interactions. An objective that economics has set for itself from that day until today, therefore, has been to design and implement institutions that would induce people to act as if they cared about those who were affected by their actions even when that was not literally true. Checkpoint 1.3: External effects a. Provide an example of a negative external effect that occurs in a social interaction. Explain why it is negative and why it is external. b. Provide an example of a positive external effect that occurs in a social interaction. Explain why it is a positive external effect. 1.4 Institutions: Games and the rules of the game Institutions Institutions are the laws, norms, and beliefs that influence how people interact, and what the outcomes of these interactions will be. People adopt the be- haviors prescribed by institutions (e.g. drive on the right if you are in the U.S.) because of some combination of • laws enforced by a government (you will be arrested and fined for driving on the left in Brazil, the U.S, France, and other countries where driving on the right is the law.) I NSTITUTIONS Institutions are the laws, informal rules, and conventions which regulate social interactions among people and between people and the biosphere. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 31 • social pressures – sometimes termed informal rules because they are not enforced by governments (your friends and neighbors will disapprove and think less of you if you drive on the left), and • information that you have about what others will do (you expect others to drive on the right, so you will avoid accidents by doing the same.) The late Nobel Laureate Douglass North called institutions the “rules of the game.”5 People can change these rules, so institutions can themselves be outcomes of games that govern how the rules of the game can be changed. Because institutions are the rules of the game for how people (and businesses, and trade unions, and governments) interact, we now introduce Game theory. Game theory uses mathematical models and verbal arguments to analyze how the outcomes of the interaction for the participants will depend on the rules of the game and the objectives of the players. It has been used extensively in economics and the other social sciences, biology, and computer science. Game theory focuses on strategic interactions where participants are interdependent and are aware of this interdependence: one player’s outcome depends on their own and other players’ actions and all players know this. We can contrast strategic with non-strategic situations in which the effect of your actions on the outcomes you will experience is independent of what others do. Your enjoyment of the program you are streaming at home alone is substantially independent of what others may be doing. But many of our economic and social interactions are strategic: G AME THEORY Game theory is the study of strategic interactions using mathematical models and verbal arguments to analyze how the outcomes of the interaction for the participants will depend on the rules of the game and the objectives of the players. E X A M P L E In 2020 under the pressure of popular protests, the government of Chile established a set of rules governing how the constitution of Chile would be amended. Another example of institutions is shown by football (soccer). FIFA governs how football can be played by what are called The Laws of the Game. These institutions also change: the corner kick was introduced in 1872 when the U.K. Football Association changed the rules. S TRATEGIC I NTERACTION An interaction is strategic when participants’ outcomes are interdependent – their well-being depends on the actions that both they and others choose, and this interdependence is known to the actors. An interaction is non-strategic when this interdependence of people’s outcomes is either absent or not recognized by the participants. A short-hand expression for the term strategic is: mutual dependence, recognized. • those considering driving to work know that their travel time will depend on how others decided to get to work that morning; • the Palanpur farmer knows that how his crop will fare if he plants early will depend on how many others planted early Checkpoint 1.4: Institutions a. What are institutions? b. What are "the rules of the game"? Games When we model strategic interactions using game theory we call the actors players. Players can be people, firms, social movements, governments and a variety of other entities. In biology, where game theory has been extensively used, even sub-individual entities are "players" such as viruses "trying to" spread or genes "trying to" get as many copies of themselves made as possible. H I S TO RY John von Neumann (1903-1957) was a Hungarian-American mathematician, computer scientist, and physicist who is regarded as the father of game theory,6 which he hoped would allow us to better understand the anti-Semitism and fascist political upheavals that he had witnessed in the early 20th century and provide the basis for understanding how groups interact. 32 MICROECONOMICS - DRAFT Players may choose from a list of possible strategies. For example, if private property is an institution that is present and enforced, then a strategy set might include "Purchase a Trek bicycle for $850." But it would not include "Pick up any available Trek bicycle," without specifying the possible penalties for stealing. The Palanpur farmers’ strategies are ‘Plant Early’ or ‘Plant Late.’ The strategies could also include a strategy based on what others did in the past (called a contingent strategy) such as: "Plant early as long as at least 5 others planted early last season." The description of a game requires us to identify the following: • Players: a list of every player in the game whether they be individuals (like the farmers in Palanpur), an organization such as Amazon or Alibaba, or some other entity that can be represented as choosing between alternative courses of action. • Strategy sets: a list for each player of every course of action available to them at each point where they must make a choice (including actions that depend on the actions taken by other players, or on chance events). The strategies selected by each of the players is called the strategy profile. S ET A set in mathematics is a collection of objects precisely defined either by enumerating the objects, or by a rule for deciding whether any particular object is in the set or not. For example, the set of positive, even integers less than or equal to 10 is, {2, 4, 6, 8, 10}. • Order of play: a game can be simultaneous such that players make their choices without knowing the choices of others, as in the game of rockpaper-scissors. Or a game can be sequential such that players move in sequence, one after the other, as in chess, so that each player knows and responds to the choices of the previous players. • Information: A game also specifies – who "knows" what, – when do they "know" it, E X A M P L E When we model the coordination problem of the Palanpur farmers as a game we assume they plant simultaneously. But when we model the interaction between a bank and a borrower we assume that the banks first makes an offer (the loan size, interest rate and schedule of repayment) and the prospective borrower responds. – is what they "know" known to others as well, – can what they "know" be used in a court of law to enforce a contract, and – is what they "know" true (this is why we use the quotation marks)? • Payoffs: Are numbers assigned to each possible outcome of the game (each strategy profile) for each player; a player chooses a strategy with the intention of bringing about the strategy profile with the highest number. It is often useful to consider payoffs as something that the players actually get. For example, considering the farmers in Palanpur again, an outcome of the game is a strategy profile indicating who plants early and late, and the payoffs could be the amount of grain each farmer harvests. We say that the payoff associated with a particular outcome of a game is how much the player values that outcome. But that means nothing more than that a player C OOPERATIVE GAME A game in which players can jointly agree upon how each will play the game (and the agreement will be respected or enforced) is a cooperative game. If no binding agreement on how to play the game is possible, then the game is non cooperative. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 33 will choose a strategy resulting in an outcome with a higher payoff number if possible. An important distinction concerning strategy sets is whether or not one of the strategies open to the players is to jointly agree on a strategy profile – that is to deliberately coordinate their actions. This is possible in what is called a cooperative game. We use the set of players, their strategy sets, their payoffs, the order of play, and the information the players have to describe the institutions governing some economic interaction, whether it is between an employer and an employee, or a central bank like the U.S. Federal reserve and a commercial bank. But even this detailed description of the interaction does not give us enough information to predict how the game will be played. The outcome of a game – how it will be played resulting in a particular strategy profile – also called a solution. To determine the solution as a way of predicting the outcome of a game we need what is called a solution concept. A solution concept for a cooperative game would include some rule for deciding on what the coordination would be, for example allowing one player selected at random to dictate the outcome, or a particular system of voting. But by positing some way that people could jointly implement some outcome, S OLUTION CONCEPT A solution concept is a rule for predicting the outcome of a game, that is, how a game will be played. cooperative game theory assumes away the problem of coordination. And the problem of how coordination is to be achieved is at the heart of the classical institutional challenge whether it takes the form of climate change or traffic jams. So we need to see how players might coordinate in what is initially a noncooperative setting – one in which coordination is not assumed at the outset – lets take a concrete example. We will use this example to illustrate a basic solution concept for non-cooperative games: the Nash equilibrium. Checkpoint 1.5: Games a. What is a game? b. How do you describe the outcome of a game? 1.5 Over-exploiting nature: Illustrating the basics of game theory People who fish for a living interact with each other regularly. Each of them are aware that how much they benefit from fishing depends not only on their own actions, but on the actions of others. This is because the more others fish, the more difficult it will be for each to catch fish. The two fishermen therefore impose negative external effects on each other, and this is why they face a coordination problem. Given that they cannot jointly decide on how Figure 1.4: Elinor Ostrom (1933-2012) was an American Political Scientist who won the Nobel Prize in economics for her work on social dilemmas, such as those encountered by Alfredo and Bob in the Fishermen’s Dilemma, and on the institutions that promote cooperation in groups. Photo Credit: Holger Motzkau. Wikimedia Commons. 34 MICROECONOMICS - DRAFT Bob 12 Hours 10 Hours Alfredo 10 Hours 12 Hours Good Worst Best Bad much to fish, each faces a basic question: how much fishing to do given the time are fishing the same waters? The game set up Specifically, we consider two fictional fishermen, Alfredo and Bob, who share access to a lake, and catch fish, which they eat. There are no other people affected by their actions. Here we illustrate the basic concepts of game theory in a game we call the Fishermen’s Dilemma. We chose the name because it is an example of what is probably the most famous game, the Prisoners’ Dilemma. As before we use a two-person example to illustrate a societal coordination problem among a much larger number of actors. The Fishermen’s Dilemma game is non-cooperative, which for two people fishing in the same lake may seem unrealistic because as neighbors they might be able to come to some kind of agreement about what each will do. We do not consider this option in the two-person case because the model illustrates a large number of people interacting. When many people interact arriving at and enforcing such a cooperative agreement would present serious challenges. Here is the game. • Players: Alfredo and Bob, two fishermen. • Strategy sets: Each may fish for either 10 or 12 hours. • Order of play : They simultaneously select a strategy, resulting in the game’s strategy profile • Payoffs: The players catch and eat an amount of fish given by the strategy profile they have implemented. This ends the game. Figure 1.5: Alfredo’s payoffs to fishing more or less depend on how much Bob fishes. Alfredo’s payoffs are described using the words like we used for the coordination problem: Planting in Palanpur. Alfredo ranks his outcomes from best to worst: Best > Good > Bad > Worst. Alfredo’s strategies and outcomes are highlighted in Blue. Bob’s strategies and outcomes are highlighted in Red (but we have not put the words to describe Bob’s outcomes in the figure). S O C I E T Y : C O O R D I N AT I O N P R O B L E M S 2 (a) A’s payoffs only 35 Bob 10 Hours 12 Hours 3 1 4 2 (b) B’s payoffs only Payoffs The payoff of each player is composed of two parts: • The amount of fish they are able to catch and consume, which they value and would like to increase; and 12 Hours 10 Hours 4 1 12 Hours 10 Hours 3 10 Hours 12 Hours Alfredo 12 Hours 10 Hours Alfredo 10 Hours Bob Alfredo Bob & ECONOMIC INSTITUTIONS 12 Hours 3 3 4 1 1 4 2 2 (c) Payoffs for both players Figure 1.6: Payoffs of players in the Fishermen’s Dilemma. Alfredo’s payoffs are in the bottom-left corner of each cell and are shaded blue. We include Alfredo’s payoffs in the right-hand and left-hand panels. Bob’s payoffs are in the top-right corner of each cell and are shaded red. We include Bob’s payoffs in the center panel and the left-hand panel. • The amount of time they spend fishing, which they find tiring and would like to decrease. We can describe the fishermen’s interaction in the form of a payoff matrix. (A matrix is a rectangular array of quantities or other quantitative information). We first present a version of the payoff matrix with words to represent Alfredo’s payoffs (but not yet Bob’s) in 1.5. Read the table this way: If Bob fishes 12 hours (the right hand column) and Alfredo fishes 10 hour (top row) this is the worst outcome for Alfredo. A payoff matrix presents hypothetical ’if-then’ information; it presents all of the possible sets of payoffs, whether or not each is likely ever to occur. The complete payoff matrix for the Fishermen’s Dilemma is represented in Figure 1.6 with numbers indicating the two fishermen’s evaluation of how good the outcome indicated is. So for example the payoff to each if they both fish ten hours (3) is fifty percent greater than if they both fish twelve hours (2). The convention we will use throughout this book is to list the row player’s payoffs first and in the bottom left corner and the column player’s payoffs second in the top right corner. So, in the Fishermen’s Dilemma game, we list Alfredo’s payoffs first and Bob’s payoffs second. We shade each players payoffs to make them easier to differentiate: blue for the row player (Alfredo) and red for the column player (Bob). Many of the games in this book involve two players and each player has two possible strategies. We often call a game like this a “2 x 2” game (a “two-bytwo” game). We now have all the elements we need for the complete description of the N ORMAL F ORM G AME We will often describe games using payoff matrices in what are called normal or strategic form, like Figure 1.10. In normal or strategic form games, we do not explicitly represent the time sequence of the actions taken by each player. We assume that each player moves without knowing the move of the other players. Normal form games therefore often represent simultaneous move games, games where players move at the same time. Simple games in normal form are often presented in a payoff matrix, a table that includes all the relevant information about the players, strategies and payoffs in the game. 36 MICROECONOMICS - DRAFT Fishermen’s Dilemma and its strategy profiles and associated payoffs. • Alfredo fishes 12 hours, Bob fishes 12 hours: When both fishermen fish 12 hours, they each catch fewer fish per hour of work, while they also have a higher cost of effort because they’ve spent a lot of time fishing. Each fisherman ends up with 2. • Alfredo fishes 10 hours, Bob fishes 10 hours: When both fishermen spend less time fishing they catch a decent amount of fish and they haven’t fished so long that the other fisherman catches fewer fish. They also benefit from a lower cost of time spent fishing. Each gets a net benefit of 3. • Alfredo fishes 10 hours, Bob fishes 12 hours: Because Bob fishes 12 hours, Al catches many fewer fish and because Bob still fishes for another two hours, he catches a lot of fish while Al doesn’t fish. Consequently, with the cost of time and catching fewer fish, Al ends up with net benefits of 1 and Bob ends up with net benefits of 4. • Alfredo fishes 12 hours, Bob fishes 10 hours: This is symmetrical to the previous description, so now Al gets net benefits of 4 and Bob gets net benefits of 1. 1.6 Predicting economic outcomes: The Nash equilibrium As you already know, to predict a game outcome – the strategy profile that will result– we need more than the description of the game alone. We need to add what is termed a solution concept – a statement about how players will behave in the game – that can be the basis of a prediction of the game’s outcome. Predicting the outcome of a game – based on the rules of the game and the solution concept – is especially important to understanding how Figure 1.7: John F. Nash (1928-2015) was an American mathematician who contributed the idea of Nash equilibrium to game theory and won the Nobel Prize in economics in 1994. His life was documented in the book and movie A Beautiful Mind.7 Source: Peter Badge, Wikimedia Commons. changing the rules of a game can change the outcome of a game. The key idea on which a solution concept is based is equilibrium. An equilibrium is a state in which there is nothing in the situation that will cause the state to change. A predicted outcome will be an equilibrium, that is, an outcome that is stationary (not changing). To understand why, imagine this were not the case. You make a prediction, but then the outcome changes. Your prediction would no longer be true because the outcome had changed. Applying this reasoning to games, if we were to predict the outcome of a game to be a strategy profile under which one or more players would have reason to change their strategy, then the prediction would be falsified as soon as they carried out the change. So the status of stationarity – change-less-ness – is a property of a prediction; and this is why equilibrium is the fundamental idea of making predictions about game outcomes. Think of a concrete example. Suppose you want to predict where a marble E QUILIBRIUM An equilibrium is situation that is stationary (unchanging) because, as long as the situation we are describing remains, there is nothing causing it to change. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 37 will be if all that you know is that it is going to be somewhere in a round bottomed salad bowl sitting on a table. If I predicted that the marble would be somewhere halfway up the side of the bowl you would doubt my prediction. The reason is that any marble in that position would move downward in the bowl, that is, its position would not be stationary so,if it ever were (for some reason) where I predicted it would be, it would not be there any longer. It is not that the prediction would necessarily be wrong. It could be true for a millisecond after I placed the marble in the bowl just above my predicted spot, for example. The only predicted position in the salad bowl that would not immediately falsify itself in this sense is the bottom. So a reasonable prediction of the location of the marble would be "the bottom of the bowl." There are some situations in which a prediction based on an equilibrium would be likely to be incorrect. Change the marble-in-bowl example by filling the bowl with very thick honey. Then if you were asked to predict where the marble would be found, you would want to know how long it had been in the bowl, did have time to reach the bottom? If if the marble had been placed in the bowl just a second ago, they you might be better off predicting that it would be where it had been placed, rather than the bottom of the bowl. The marble-in-bowl-of-honey is often a better illustration of how economic processes work than the initial example. Markets are often out of equilibrium. Predicting things in motion is a much more challenging task than predicting them when they are stationary. We provide an example in our model of residential the segregation (below) where we are able to follow the process of change step by step. But for the most part we study equilibria and how to change them so as to improve outcomes. In the marble-in-bowl illustration (without the honey) what is the solution concept that let us arrive the "bottom of the bowl" prediction? It is gravity, which is our understanding about a reasonable way for the marble to "behave." In modeling an economic interaction, the game structure is analogous to the salad bowl. What is the analogy to gravity? The answer is the players’ best response. Best-response strategies By far the most widely-used solution concept, the Nash equilibrium, is based B EST RESPONSE A player’s best response is a strategy that results in the highest payoff given the strategies of the other players. on the idea that players choose best-response strategies; they do the best they can given the strategies adopted by everyone else Bob To understand better what a best response is, think about Alfredo’s choices in 10 Hours sponses. First, what strategy should Alfredo adopt in order to gain the highest Alfredo for finding the Nash equilibrium on the basis of your analysis of his best re- 12 Hours 10 Hours the Fisherman’s Dilemma. We will also introduce the "circle and dot" method 12 Hours 3 3 1 ● 4 4 1 ● 2 2 Figure 1.8: Hold constant Bob playing Fish 10 hours to assess Alfredo’s best response to Bob’s playing Fish 12 hours. 38 MICROECONOMICS - DRAFT payoff if Bob were hypothetically to play Fish 10 hours (we say "holding constant" this strategy) as shown in Figure 1.8. • Against Bob playing Fish 10 hours, Alfredo can get a payoff of 3 for fishing 10 hours or a payoff of 4 for Fishing 12 hours. • 4 > 3 therefore fish 12 hours is Alfredo’s best response to Bob playing fish Bob 10 hours. 10 Hours Let’s repeat analysis and hold constant Bob playing Fish 12 hours, as shown in Figure 1.9. • Against Bob playing Fish 12 hours, Alfredo can get a payoff of 1 for playing Fish 10 hours or a payoff of 2 for playing Fish 12 hours. 12 Hours 10 Hours 10 hours) to indicate that it is Alfredo’s best response. Alfredo • place a solid point in the cell (Alfredo plays Fish 12 Hours, if Bob plays Fish 12 Hours 3 3 ● 4 4 1 1 ● 2 2 Figure 1.9: Hold constant Bob playing Fish 12 hours to assess Alfredo’s best response. • 2 > 1 therefore Fish 12 hours is Alfredo’s best response to Bob playing fish 10 hours. • place a solid point in the cell (Alfredo plays Fish 12 Hours, Bob plays Fish 12 hours) to indicate that it is Alfredo’s best response. Checkpoint 1.6: A best response for Bob Repeat the process we went through for Alfredo, but do it for Bob instead. Notice that when you do so, you will blank out a row for Alfredo to hold his strategy constant, whereas you blanked out a column for Bob to hold his strategy constant. What are Bob’s best responses? Show his best responses using a hollow circle. M CHECK: STRONG AND WEAK BEST R E S P O N S E . A best response may be either strong or weak. A strong (also called strict) best response yields higher payoffs than any other: it is strictly "better" than any other strategy. There can be no strategy that is better than a weak best response but a weak best response need not be better than any other; it may be "as good as" (the payoffs to the strategy and some alternative strategy being equal.) Nash Equilibrium and the outcome of a game Using the best responses of the players we can now predict the outcome of a game using as our solution concept the Nash equilibrium. A Nash equilibrium is a profile of strategies – one for each player – each of which is a best response to the strategies of the other players. A Nash equilibrium is also called a mutual best response. Because at a Nash equilibrium all players are playing their best response to all of the others, it follows that no player has a reason to change his or her strategy as long as the other players do not change theirs. Some games do not have a Nash equilibrium and you will see shortly that some have more than one. In Figure 1.10, Alfredo’s best responses are shown by the solid black dot in the cell. Bob’s best responses are shown by the hollow circle. Their best responses coincide at the Nash equilibrium (Fish 12 Hours, Fish 12 Hours) with payoffs (2, 2) shown in the cell where the solid point is inside the hollow circle. You can use the "dot and circle" method to find one or more Nash E X A M P L E The Rock,Paper, Scissors game (also called ro-sham-bo and by many other names in other languages) originated in China about two thousand years ago. It does not have a Nash equilibrium. N ASH EQUILIBRUM A Nash equilibrium is a profile of strategies – one strategy for each player – each of which is a best response to the strategies of the other players. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S 12 Hours 10 Hours Alfredo 12 Hours 3 3 4 4 1 ● 1 39 Figure 1.10: Payoff Matrix for The Fishermen’s Dilemma. The solid dots indicate Alfredo’s best responses. The hollow circles indicate Bob’s best responses. A Nash equilibrium is a cell that contains both dots. In this case there is just one Nash equilibrium: both fishing 12 hours. Bob 10 Hours & ECONOMIC INSTITUTIONS ● 2 2 equilibria (if they exist) for games that can be represented by a payoff matrix like Figure 1.10. In the Fishermen’s Dilemma game described by Figure 1.10. Each player’s best response to both of the others’ strategies was Fish 12 hours. Therefore only one outcome is a Nash equilibrium: both fishermen fishing 12 hours. We say that (Fish 12 Hours, Fish 12 Hours) is the Nash equilibrium with payoffs (2, 2). The outcome demonstrates how Nash equilibrium can initially seem counterintuitive. Both would have had higher payoffs if they could have agreed to restrict their fishing to 10 hours (they could have had 3 each if they both fished 10 hours and 3 > 2). But suppose both were for restricting their fishing to 10 hours; then both would an incentive to fish for 12 hours (because 4 > 3) and unless they had a binding agreement to continue fishing less, both would choose to fish more. The Fishermen’s Dilemma is therefore a coordination problem and it returns us to the classical institutional challenge. Without institutions to align the individual interest of the participants with their shared interest, they get an outcome that is worse for both of them than other possible outcomes. We will later show how a change in the institutions reglating how Alfredo and Bob interact – that is, changing the rules of the game – might address this coordination. Checkpoint 1.7: Nash Equilibrium a. Explain why none of the other three outcomes (those that are not, (Fish 12 Hours, Fish 12 Hour) of the Fishermen’s Dilemma satisfy the definition of Nash equilibrium. b. At each of the other three outcomes, which player has an incentive to change strategy and in what way? Explain. c. Explain why a game like Rock Paper Scissors would not be much fun if there was a Nash equilibrium. 40 MICROECONOMICS - DRAFT Dominant strategies In the Fisherman’s Dilemma (and all Prisoners’ Dilemmas) there is a single strategy that yields the highest payoffs to a player for both (or all of if there are more than two) of the strategies that the other player might adopt. A strategy is a player’s dominant strategy if it is the player’s best response to all possible strategy profiles of the other players. That is, a strategy is a dominant if by playing it the player’s payoff is greater than or equal to the payoff playing any other strategy for every one of the other player’s profiles of strategies. Likewise we say that strategy A is dominated by another strategy B if the payoff to playing B is at least as great or greater than playing A for every strategy profile of the other players. If there is a strategy that dominates all of the other strategies that an player may choose, then it is a dominant strategy. If each player in a game has a dominant strategy, then the strategy profile in D OMINANT STRATEGY A strategy is dominant if by playing it the player’s payoff is greater than or equal to the payoff playing any other strategy for every one of the other players profiles of strategies. A strategy is dominant if it is the player’s best response to all possible strategy profiles of the other players. A dominant strategy dominates all of the other strategies available to the player. which all players adopt their dominant strategy is called a dominant strategy equilibrium. We can apply the concept of dominant strategy equilibrium to the Fishermen’s Dilemma. To do so, we need to understand whether each player has a dominant strategy. • When Alfredo fishes 10 hours, his payoff is 3 if Bob fishes 10 hours and 1 if Bob fishes 12 hours. • When Alfredo fishes 12 hours, his payoff payoff is 4 when Bob fishes 10 hours and 2 when Bob fishes 12 hours. • So, when Bob fishes 10 hours, fishing 12 hours gets Alfredo a higher payoff (4 > 3) and when Bob fishes 12 hours, fishing 12 hours gets Alfredo a higher payoff (2 > 1) • Therefore, Alfredo gets a higher payoff from fishing 12 hours against each of his opponent’s strategies • Fish 12 hours is therefore Alfredo’s dominant strategy. Fishing 12 hours is also Bob’s dominant strategy. Because each player has a dominant strategy to fish 12 hours, the dominant strategy equilibrium is (Fish 12 hours, Fish 12 hours) with payoffs (2, 2). The fact that the Fishermen’s Dilemma has a dominant strategy equilibrium makes it a particularly simple problem (both for us, studying it, and for the players because what is best for each does not depend on what the other does). The dominant strategy equilibrium of a game is always a Nash equilibrium, so in the Fishermen’s Dilemma, the outcome where both players fish 12 hours is the only Nash equilibrium. D OMINANT STRATEGY EQUILIBRIUM . A dominant strategy equilibrium is a strategy profile in which all players play a dominant strategy. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S Checkpoint 1.8: Dominance and Nash Equilibrium a. Repeat the analysis we did for Alfredo for Bob and confirm that 12 hours is a dominant strategy for him too. b. We said that a dominant strategy equilibrium is always a Nash equilibrium. But do you think that a Nash equilibrium is always a Dominant Strategy equilibrium? Why or what not? 1.7 Evaluating outcomes: Pareto-comparisons and Pareto-efficiency The Nash equilibrium can help us predict the result of a particular interaction. But it does not tell us anything about whether some outcome is good by any standard, or even better or worse than some other outcome. Economists, policy-makers and others would like to evaluate whether some outcomes are better or worse so that we can try to work out which rules of the game would make the better outcomes Nash equilibria, and therefore more likely to be what we observe. In Chapter 16 we show how economics deals with this for questions of public policy. The challenge in making these comparisons is that whether some outcome is better than another depends on what you value, and there is no agreed upon standard of what makes one outcome better than another. Returning to our fishermen, here are some of the values that one could use in evaluating an outcome • Fairness in the distribution of payoffs among the players; is it fair that Alfredo receives 4 times what Bob gets when Alfredo does not limit his fishing hours and Bob does? • Are the rules of the game itself fair? In the Fishermen’s Dilemma the same rules applied to both players; but were the game a bit different, many would think it unfair if Alfredo could simply order Bob to fish 10 hours, or to hand over half of all the fish Bob caught. • Setting aside fairness, is the outcome a reasonable use of available resources including the working time of the two fishermen and the sustainability of the lake itself and the living things that it supports. There are many other standards that could be proposed. Questions of better and worse are called "normative question." With normative questions matters of ethics or morals are necessarily involved. We will introduce some experimental evidence on people’s views concerning fairness in Chapter 2 and some analytical tools for studying normative issues in Chapter 13. Here we introduce a concept that economists use to evaluate economic outcomes. The idea is simple: an objective of public policy and institutional design – the rules of the game – should be to avoid those outcomes – like traffic & ECONOMIC INSTITUTIONS 41 42 MICROECONOMICS - DRAFT 6 Bob 12 Hours 10 Hours Alfredo 10 Hours 12 Hours 3 3 c b 4 d 3 c 2 a 4 1 1 4 Bob's payoffs 5 d 2 2 a 1 b 0 0 (a) Fishermen’s Dilemma with Labeled Points 1 3 4 5 6 Alfredo's payoffs (b) Analyzing points for Pareto efficiency jams, planting late in Palanpur, and over-fishing the lake – that are worse for everyone, compared to an alternative outcome that also would have been feasible. Pareto comparisons To compare outcomes when more than one person is involved economists use the concept of Pareto-efficiency based on Pareto-comparisons of outcomes. An outcome is Pareto superior to another if it allows at least one of those involved to be better off without anyone being worse off. A Pareto-superior outcome is also called a Pareto improvement over the outcome it was compared to. This is a Pareto comparison. An outcome is Pareto efficient if no other feasible outcome is Pareto superior to it. Figure 1.11 depicts the outcomes of the interaction between Alfredo and Bob. Alfredo’s payoffs are on the horizontal axis (x-axis), so the outcomes get better for Alfredo as you move from the left to the right. Bob’s payoffs are on the vertical axis (y-axis), so the outcomes get better for Bob as you move from bottom to top. The left panel of Figure 1.11 is the Fishermen’s Dilemma payoff matrix with each outcome given a label a, b, c, or d. These same points appear in the right panel of Figure 1.11 where you can read on the vertical and horizontal axes the payoffs to the two players that you see in the payoff matrix. The Pareto-comparison is geometrically easy to see in this type of plot. An outcome is Pareto-superior to another if the first outcome lies to the "northeast" of the second in the plot. "North-east" in this figure is "better for both." An outcome is Pareto-efficient if there is no other feasible outcome to the 2 Figure 1.11: Three Pareto-efficient outcomes of the Fishermen’s Dilemma. In 1.11 a, the outcomes are labeled in the bottom right corner as a for (2, 2), b for (4, 1), c for (3, 3) and d for (1, 4). These allow us to make Pareto comparisons for the outcomes of the Fishermens Dilemma. Alfredo’s payoffs are plotted on the horizontal axis, increasing as you move rightward. Bob’s payoffs are plotted on the vertical axis, increasing as you move upward. In 1.11 b we show the Fishermen’s Dilemma indicated by the four possible outcomes given by the same letters that appear in each of the cells of the payoff matrix. We use shaded colors indicating 90 degree angles to the northeast of the feasible outcomes (each of the lettered points). S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 43 north-east. Or if you think of the colored areas whose lower left corners are points a, b, c, and d then a Pareto efficient point is one that has no other point in its "colored shadow" extending upwards and to the right of the point. When two different outcomes are both Pareto-efficient, they cannot be Paretocompared or Pareto-ranked. We could rank c above a because both players were better off, but with b, c and d we cannot move from one outcome to another without worsening outcomes for at least one of the players. PARETO PARETO An outcome is Pareto superior to another if it allows at least one of those involved to be better off without anyone being worse off. A Pareto-superior outcome is also called a Pareto improvement over the outcome it was compared to. This is a Pareto comparison. An outcome is Pareto efficient if no other feasible outcome is Pareto superior to it. If we can rank two outcomes such that one is Pareto superior to the other, then we say that these two outcomes can be Pareto compared, or Pareto ranked. COMPARISONS AND EFFICIENCY Here is a checklist to use when evaluating any given set of payoffs for both players. Consider a point called x with payoffs pxA for Player A and pxB for player B and compare it to another theoretical point, y with its corresponding payoffs: • For any point, x, check whether there exists an alternative point y where at least one player gets a payoff that is greater than pxA or pxB without the other player being worse off. • If at y, pyA > pxA while pyB pxB or pyB > pxB while pyA pxB then y is Paretosuperior to x (at least one player is better off while the other player is not worse off). • If no other point exists where at least one player is better off with no other player being worse off (no Pareto-superior point exists), then x is Paretoefficient. Checkpoint 1.9: Pareto efficiency and Pareto improvements in the Fishermen’s Dilemma Referring to Figure 1.11, consider the following: a. Is any point dominated by some other point? Say which, if any? item At which point is the total payoff of the two fishermen the greatest? b. Would a change from any other point to that "total payoff maximum" point be a Pareto improvement? 1.8 Strengths and shortcomings of Pareto efficiency as an evaluation of outcomes Pareto efficiency gives us a way to identify "lose-lose" outcomes we should seek to avoid, namely those " that are worse for all than they could be" . But, as we will now see, except in special cases, Pareto efficiency does not provide a rule to select what we might call in everyday speech "the best" outcome. To see why this is true, suppose we have a cake of given size and we are dividing it among people, all of whom equally enjoy eating cake. An outcome in which one person gets the entire cake is surely Pareto-efficient because in any other allocation that lucky person would get less. Likewise an E X A M P L E We will see in Chapter 2 that people often reject a highly unequal division of some "pie" and are willing to sacrifice a substantial amount of money rather than to accept what they consider to be an unfair division. And we will see in Chapter 13 how people’s desires for a fair society might allow for a choice among outcomes that differ in who gets what. 44 MICROECONOMICS - DRAFT allocation in which everyone got an equal slice of the cake is Pareto-efficient, for in any other allocation at least one person would have to get less. The Pareto criterion can say nothing about such distributional fairness. All it says is "make sure there’s no cake left on the table!" When there are many Pareto-efficient outcomes there is always a conflict of interest among players over which outcome they would prefer we cannot say that one is "more Pareto-efficient" than the other. It is also perfectly sensible to prefer an outcome that is not Pareto efficient but is more fair over an alternative Pareto efficient outcome that is unfair. To continue the cake example, if there are two people between whom the cake will be divided many people would reject the (Pareto efficient) outcome in which one person gets the entire cake in favor of a Pareto inefficient alternative in which each gets a third of the cake (the remaining third perhaps being thrown away or destroyed in the conflict over its distribution). But the Pareto comparison would remind us that each person getting half of the cake is preferable to each getting a third with the rest being wasted. Pareto efficiency is a particular device for screening out those outcomes (like throwing away some of the cake in the above example, or planting late in Palanpur) that should not be among the list of candidate feasible outcomes among which the choice of better or best should be made or grounds of fairness or other bases. Other ways of "screening" the list of candidate outcomes would give priority not to individuals’ payoffs but to whether the rules of the games that produced the outcomes are themselves fair and consistent with other values such as individual dignity, respect and freedom. Checkpoint 1.10: Pareto efficiency Consider these questions about Pareto efficiency. a. True or False (and explain): "The fact that an outcome is Pareto-efficient does not imply that it is preferred by all the actors to all the other outcomes." b. Can two Pareto efficient outcomes be Pareto compared? Why or why not? Explain. c. Imagine you are an impartial observer evaluating the possible outcomes that might occur for Bob and Alfredo. Are there any reasons why you might judge the Pareto-inefficient outcome a in the figure to be better than the Pareto efficient outcomes b and d, despite the fact that a is Pareto-inefficient? 1.9 Conflict and common interest in a Prisoners’ Dilemma The game the fishermen are playing is a particular case of the Prisoners’ Dilemma. A Prisoners’ Dilemma is a two-person interaction in which there is P RISONERS ’ D ILEMMA A Prisoners’ Dilemma is a two-person social interaction in which there is a unique Nash equilibrium (that is also a dominant strategy equilibrium), but there is another outcome that gives a higher payoff to both players, so that the Nash equilibrium is not Pareto-efficient. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S 10 Hours (Defect) (Cooperate) 12 Hours Alfredo 12 Hours (Defect) y w w x x y 45 Figure 1.12: A General Prisoners’ Dilemma. For the game to be a Prisoners’ Dilemma, we require y > w > z > x and 2w > y + x (this is like 4 > 3 > 2 > 1 and 2 ⇥ 3 > 4 + 1 from the numerical example). Bob 10 Hours (Cooperate) & ECONOMIC INSTITUTIONS z z a unique Nash equilibrium (that is also a dominant strategy equilibrium), but there is another outcome that gives a higher payoff to both players, so that the Nash equilibrium is not Pareto-efficient. So, in the Prisoners’ Dilemma both players get their second worst payoffs in the game by playing their strictly dominant best-response strategies. We now point out some of the general characteristics of this particular kind of coordination problem. To do this, in Figure 1.12 we show the familiar payoff matrix for the Fishermen’s Dilemma, but instead of the numbers indicating the payoffs of the players now we label the payoffs w, x, y, and z. We label the action of fishing 10 hours "Cooperate" because it is the action the two fishermen would take if they could coordinate their actions. The action of fishing 12 hours is labeled "Defect" because a fisherman who chooses to fish 12 hours is deviating from the mutual cooperate outcome on which the two fishermen might be able to coordinate. The interaction is a Prisoners’ Dilemma if two conditions hold: • y > w and z > x means that fishing Defect is a strict dominant strategy • w > z means that mutual cooperation is Pareto superior to mutual defection. For Alfredo, 12 Hours is a best response to Bob playing 10 Hours because y > w; 12 Hours is also a best response to 12 Hours because z > x (both best responses are shown by the solid point). Similarly, for Bob, 12 Hours is a best response to Alfredo playing 10 Hours because y > w; 12 Hours is also a best response to 12 Hours because z > x (both best responses are shown by the hollow circle). Therefore the Nash equilibrium is (12 Hours, 12 Hours) with payoffs (z, z). If the players play the game non-cooperatively (they do not coordinate their actions) each will play their dominant strategy – defect – and get z when by cooperating they could have each received w. M AT H N OT E A third condition is sometimes added, namely x + y < 2w which means that the sum of payoffs when both players cooperate is greater than the sum of payoffs when one cooperates and the other defects. This condition makes (Cooperate, Cooperate) preferable to any outcome in which one defects and the other cooperates. 46 MICROECONOMICS - DRAFT Economic rent: The incentive to coordinate Both players have a good reason to try to change the rules of the game so that they can agree on both cooperating. How much more they would get if they were to mutually cooperate than if they mutually defected – in this case w z – is called an economic rent, meaning the difference between the payoff that they would get if they cooperated and their next best alternative. Their next best alternative to cooperating, we assume, is mutual defection, also known as their fallback position. FALLBACK POSITION A player’s fallback position is the payoff they receive in their next best alternative. Economic rents and the fallback position play a central role in microeconomic theory, so it is a good idea to master them. The meaning of fallback position is intuitive: it is what you fall back to if your current outcome is not possible, in this case if the mutual cooperation should not work out. A player’s fallback position is the payoff they receive in their next best alternative. The term "economic rent" may at first seem surprising, because the word "rent" also means a payment for the temporary use of something like a rent paid to a landlord or a rent or a car rental agency. The term economic rent means something entirely different. A participant’s economic rent is the payoff they receive in excess of what they would get in their fallback position. We shall use the idea of a fallback often, from social interactions like the Prisoners’ Dilemma, to worker-employer relationships where a worker wants a job more than being unemployed, to a person applying to a bank for a loan rather than trying to get money from friends, family, or the government. As these examples indicate, in real life the fallback position will differ depending E CONOMIC RENT A participant’s economic rent is the payoff they receive in excess of what they would get in their fallback position. When we use the term " rent" we mean economic rent. The sum of the economic rents received by the participants in an interaction is sometimes termed the economic surplus. on the details of the situation that we set aside when we model interactions like fishing in a lake, employment and borrowing. Impediments to coordination: Limited information and conflicts of interest If w z is substantial – meaning substantial rents associated with cooperation for each player – then it might seem a simple matter for the players to agree to cooperate. But people often fail to reach or enforce such an agreement, for two main reasons: • Limited information. The participants may lack the information needed to monitor and enforce an agreement. How can a participant know or verify what other participants do? • Conflict over distribution of the economic rents from cooperation Disagreement about who gets what – for example who gets to fish more – may make it impossible for the two to agree. Concerning the information problem, the fishermen, for example, may have no way of enforcing an agreement, or even knowing if the agreement has been violated. While each may know how many hours the other has fished on day E X A M P L E The term economic rent is what is known in the study of language as a “false friend,” a term that you think you know the meaning of but mean something entirely different in the new language you are now learning. "Sensible" in English means "reasonable" but in Italian it means "sensitive." S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS with clear and sunny weather, on a foggy day it may be impossible to know. Even if one fisherman knows how much the other fished, that knowledge may be insufficient to enforce an agreement through a third party such as a court of law. This is the problem of asymmetric information or non-verifiable information. Information is asymmetric if people know different things, or if what one person knows (for example how many hours he fished), the other person does not know. Information is not verifiable if people cannot use it to enforce an agreement or a contract. For example most courts will not accept "hearsay" (meaning "second hand") information, so if one of the fishermen had heard from someone else that the other had fished 12 hours, this would be nonverifiable information. Asymmetric and non-verifiable information will play a central role in our analysis of how the labor market, the credit market and other markets work. Concerning the conflicts over the distribution of the economic rents from cooperation, the Fishermen’s Dilemma, the agreement to restrict fishing to 10 hours a day, is an agreement both to restrict fishing and to divide the benefits of restricting fishing in a particular way, namely equally. But the fishermen need not agree on 10 hours each. Alfredo might insist that he will fish 12 hours and Bob only 10 hours. Or Bob might insist on the opposite. Or Bob might insist that both fish 10 hours, but that Alfredo give him most of Alfredo’s catch, leaving Alfredo with just enough of his catch to be no worse off than had they both fished 12 hours, namely with a payoff of z. Unless they can find a mutually acceptable solution to the distribution problem they may end up having no agreement at all, and then simply fish at 12 hours each, at their fallback position. The fishermen’s distribution conflict highlights a challenge that arises in any voluntary economic interaction. Consider their possible agreement to limit their fishing time: • The agreement is voluntarily entered into. This means that neither player can force the other to accept terms worse than their fallback position. • The agreement therefore must allow each participant to achieve a payoff greater than (or at least not worse than) had the individual not agreed to cooperate. In other words, there must be some economic rents made possible by a voluntary cooperative outcome. • This being the case, the participants have to find a way that the total rents will be divided.. If they are to agree to cooperate by restricting fishing, they must also agree on how these economic rents will be distributed. • Conflict over the distribution of the economic rents (who gets what amount ASYMMETRIC INFORMATION Information is asymmetric if something that is known by one participant is not known by another. N ON - VERIFIABLE INFORMATION cannot be used to enforce a contract or other agreement 47 48 MICROECONOMICS - DRAFT of economic rent) may prevent the fishermen from coming to an agreement. We sometimes think of cooperation and conflict as opposites, as for example when members of a team cooperate in their efforts to win some conflict with another team. But the Prisoners’ Dilemma is a scenario of conflict and cooperation among the very same participants. They have common interests in getting some share of the economic rents by cooperating; but they have conflicting interests in how the total will be divided into the rents received by each. A catalogue of games: And their challenges to coordination Some interactions present greater impediments to coordination than other; the Prisoners’ Dilemma is in some respects the most challenging of all. We can classify coordination problems and the challenges they present by the relation between Nash equilibria and Pareto-efficient outcomes of the games that represent them. • In the Prisoners’ Dilemma, you know, there is a unique Nash equilibrium that is Pareto-inefficient. Because this outcome is also a dominant strategy equilibrium, coordination on mutual cooperation will not occur even if one of the players insists (perhaps for moral reasons) on cooperating. • In interactions like Planting in Palanpur, which are often called Assurance Games, there are two Nash equilibria, (both Plant Early and both Plant Late) one of which (Plant Early) is Pareto-superior to the other (Plant Late). In these games if one of the players plays the strategy making up the Pareto superior equilibrum (Plant Early) then the best response of the other will be to do the same. Finding institutions that will implement the preferred plant early outcome in a game like this will be a lot less challenging than in a Prisoners Dilemma. • Another important class of coordination problems arise in what we call Disagreement Games where there are two Nash equilibria each of which is Pareto-efficient, so that they cannot be Pareto-ranked, and players disagree about which Nash equilibrium they would like occur. These are like the Planting in Palanpur game but with the additional challenge stemming from a conflict over which Nash equilibrium will be implemented. We start with an even less challenging game in which players’ self interests lead them to a Pareto-efficient Nash equilibrium. F AC T C H E C K In the next chapter we will see that across many cultures of the world, people would rather get nothing than get what they consider to be an unfair share of the economic rents, and as a result cooperation breaks down and nobody gets any rent at all. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S Barbara Tomatoes 1.10 Corn Tomatoes Arkady Corn 2.5 1 49 Figure 1.13: An Invisible Hand Game with best responses indicated by circles and dots. Arkady’s payoffs are listed first in the bottom-left corner. Barbara’s are listed second in the top-right corner. The game captures Adam Smith’s ideas of specialization and gains from trade (that is, the opportunity to obtain economic rents from trade). 2 2 ● 4 & ECONOMIC INSTITUTIONS 4 ● 1 2.5 Coordination successes: An invisible hand game The characteristic of an invisible hand game is that it has a Nash equilibrium that is Pareto-efficient. The Invisible Hand game illustrates Adam Smith’s core insight that through the competitive buying and selling of privately owned goods on competitive markets, self-interested people can achieve outcomes to the benefit of all at least some conditions (that we spell out in Chapter 14). In modern economic language, we would say they avoid Pareto-inefficient outcomes. Though our game is much simpler than Smith’s reasoning and Smith did not use ideas like Pareto efficiency, our game illustrates an aspect of Adam Smith’s idea of how the invisible hand works. The participants, pursuing their self-interest, reach an outcome that beneficial for all of them. (We return to how the invisible hand is understood in contemporary economics in Chapter 14). Consider a 2-by-2 game between two players, Arkady and Barbara, who are both farmers. Each player can choose between two strategies: planting corn or planting tomatoes. The payoffs that they achieve are provided in the payoff matrix in Figure 1.13, which we call the Corn-Tomatoes game. The payoff matrix reflects two facts about the problem that the two farmers face. • Either because of their skills or the nature of the land they own, Arkady is better at growing tomatoes; Barbara is better at growing corn • They both do poorly when they produce the same crop because the increased supply of whichever good it is that they both produce drives down the price. The Nash equilibrium of the Corn-Tomatoes game is (Tomatoes, Corn), that is, Arkady plants tomatoes, and Barbara plants corn, at which the players receive payoffs (4,4). (Tomatoes, Corn) is Pareto efficient as there is no alternative I NVISIBLE H AND G AME In an invisible hand game there is a Nash equilibrium that is Pareto-efficient. 50 MICROECONOMICS - DRAFT 5 Bina Late 3 4 4 c 0 Pareto−efficient Nash equilibrium c 4 Bina's payoffs Early Late Aram Early A's fallback b 0 2 d a 3 b a 2 Pareto−inferior Nash equilibrium 1 d 0 3 2 0 (a) Assurance Game with labeled payoffs 1 3 4 5 (b) Payoffs plotted against each other Just as in Adam Smith’s reasoning about his invisible hand, Arkady and Barbara, in their interaction, through competition with each other and following their self-interest coordinate their economy to their mutual benefit. In the Invisible Hand Game, each player pursues self-interested objectives and benefits from the fact that the other does too. In an Invisible Hand game individual incentives lead people to act in ways that promotes mutual benefit. Checkpoint 1.11: Invisible Hand Game Which entries in the payoff matrix would you have to compare in order to show the follwing: a. They each do better when Arkady specializes in tomatoes and Barbara specializes in corn then vice versa. b. They each do worse when both produce the same crop. c. Growing corn is Barbara’s dominant strategy d. Arkady growing tomatoes and Barbara growing corn is the dominant strategy equilibrium. e. Explain why the Nash equilibrium of the game is Pareto efficient. Assurance Games: Win-win and lose-lose equilibria Return to the farmers in Palanpur. There are two Nash equilibria in this game, one in which both participants Plant Early and one in which both Plant Late. The best response to the other farmer planting early is also to plant early, while the best response to the other farmer planting late is also to plant late. The outcome where both farmers plant early is Pareto-superior to the outcome when both farmers plant late. 2 Aram's payoffs outcome which is Pareto superior to it. 1.11 B's fallback Figure 1.14: Planting in Palanpur: An Assurance Game. Aram’s payoffs are listed first in the bottom-left corner. Bina’s payoffs are listed second in the top-right corner. Aram’s best responses are shown by the solid point and Bina’s are shown by the hollow circle. The Nash equilibria of the game are (Plant Early, Plant Early) and (Plant Late, Plant Late), with payoffs (4, 4) and (2, 2). The Plant Early Nash equilibrium is Pareto-efficient. The Plant Late equilibrium is not. In the right-hand panel, the payoffs are plotted against each other. Aram’s payoffs are plotted on the horizontal axis, increasing as you move rightward. Bina’s payoffs are plotted on the vertical axis, increasing as you move upward. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 51 The players do not have any conflict of interest: both would share equally in the gains to cooperation, should they find a way to coordinate on planting early. The problem for the real life farmers of that village is that they are stuck in the Pareto-inefficient Nash equilibrium of an Assurance Game. Their challenge is how move to the Pareto-superior Nash equilibrium. This could happen if all the participants had confidence (were assured) that the other participants would follow their lead in moving to the superior outcome. This is why this type of game is often labeled as an "Assurance Game." Figure 1.14 is the payoff matrix for two players, Aram and Bina, choosing A SSURANCE G AME In an Assurance Game, there are two Nash equilibria, one of which is Pareto-superior to the other. The Planting in Palanpur Game is an example. when to plant their millet in the village of Palanpur, India. (It is the same as the earlier figure about the two farmers, except that we now have numbers representing the farmers’ payoffs). Coordination failures arise in the Assurance Game because of positive feedbacks: the more people who plant late the more is the incentive for others to plant late, and vice versa. Each strategy exhibits strategic complementarity. Checkpoint 1.12: Graphing Palanpur Using the graphical method for identifying Pareto-efficient outcomes as shown in Figure 1.14, show which outcomes in the Palanpur game are Pareto-efficient. Can you explain why a and c are Nash equilibria? Assurance game and strategic complementarity Social media, dating platforms, and other matching services are examples of strategic complementarities. They are more valuable to for everyone if many people participate. Strategic complementarity exists when either of two conditions hold. 1. A strategy is a strategic complement to itself : The payoff to playing a particular strategy increases as more people adopt that strategy as a result of some form of positive feedbacks. Dating platforms are an example. The strategy could be “Open a Tinder Account.” The positive feedback arises because the more other people that are using Tinder the more people you will “meet.” Plant Early in Palanpur is another example as we will see. 2. One strategy and another are strategic complements to each other. The payoff to playing one strategy (say, A) is greater the more people adopt the other (B) in which case we say that strategies A and B are strategic complements. An example is the Invisible Hand Game shown in Figure 1.13. The payoff to Arkady from planting tomatoes is greater if Barbara plants corn (instead of tomatoes), and the payoff to Barbara from planting S TRATEGIC C OMPLEMENTARITY Strategic complementarity exists when i) A strategy is a strategic complement to itself : The payoff to playing a particular strategy increases as more people adopt that strategy as a result of some form of positive feedbacks, or ii) One strategy and another are strategic complements to each other. The payoff to playing one strategy (say, A) is greater the more people adopt the other (B) in which case we say that strategies A and B are strategic complements. 52 MICROECONOMICS - DRAFT corn is greater if Arkady plants potatoes (instead of corn). Growing corn and growing tomatoes are strategic complements. The farming in Palanpur problem is a case of the first, not the second. But nobody has any reason to participate if no others do. This is an example of what are called network externalities or network external effects which occur when the benefits to members of a social or physical network increase when more people join the network. One example is a particular social network: if you’re the only person on it, there is really no point. But, as more and more people join the network, then social networking site becomes more useful and your payoff increases with the number of users in the network. These so-called network externalities are a particular case of strategic complementarity. By joining a network each person confers an external benefit on the existing members of the network. We predict and evaluate the possible outcomes of the Planting in Palanpur Game using the concept of best response (using the dot and circle method introduced earlier). And we see that the game has two Nash equilibria (Early, Early) with payoffs (4, 4) and (Late, Late) with payoffs (2, 2). (Early, Early) is Pareto-superior to (Late, Late) and it is Pareto-efficient because no alternative outcome is Pareto-superior to (Early, Early). Even though there is a Pareto efficient Nash equilibrium, a population – like the people of Palanpur – may get stuck in the Pareto inferior Nash equilibrium. that does not guarantee players will actually play it. So we have two conclusions: • The fact that a Pareto efficient outcome is a Nash equilibrium does not mean that it will be the one we observe and • In cases where there is more than one Nash equilbrium, we need more information than is provided by the Nash equilibrium and Pareto efficiency concepts to make a prediction about the strategy profiles we will see in practice. Checkpoint 1.13: Assurance Game Which payoff table entries would you have to compare in order to show that: a. Planting early is Pareto efficient. b. Planting late is a Nash equilibrium. c. The best response to the other planting early is to plant early. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S 1.12 Improve Stick to English Swahili Aisha Ben Improve Swahili Stick to English 2 0 ● 0 4 1 1 ● 4 2 Disagreement Games: Conflict about how to coordinate We use the Language Game described in Figure 1.15 as an example of a Disagreement Game. A Disagreement Game illustrates coordination games in which there are two Pareto-efficient Nash equilibria (which are therefore Pareto-incomparable), and the players are in conflict over which Nash equilibrium each prefers. So, the players’ problem is to manage to coordinate on one of the Nash equilibria, or alternate systematically between them, to ensure that no coordination failure results and they do not end up at an outcome neither would prefer. Consider two players, a home-language Swahili-speaker (Aisha) and a homelanguage English-speaker (Ben) who have recently met. Each person can speak the other language, but prefers to speak their home language. They share many common interests but do not communicate as well as they would like. Each has two strategies: Stick to your home language or Improve the other language. Among the possible outcomes are that he could learn better Swahili and they could routinely converse in that language; and she could learn better English and they could converse in English. They do not need to both be fluent in both languages. So for Aisha, if Ben becomes fluent in Swahili, then her best response is not to take the time and trouble to improve her English. For Ben, similarly, if Aisha were to become fluent in English, then there would be little point in taking the Swahili courses. The result is two Nash equilibria (Stick to Swahili, Improve Swahili) with payoffs (4, 2) and (Improve English, Stick to English) with payoffs (2, 4). The two Nash equilibria are both Pareto-efficient because there are no alternative outcomes which are Pareto-superior to the Nash equilibria. But, as shown in the payoff matrix 1.15, Aisha would prefer the (Stick to Swahili, Improve Swahili) Nash equilibrium and Ben would prefer the (Improve English, Stick to English) Nash equilibrium. & ECONOMIC INSTITUTIONS 53 Figure 1.15: A Disagreement Game: The players need to coordinate on an equilibrium, but each prefers one equilibrium to the other, so there is a conflict of interest. If they fail to coordinate on one of the Nash equilibria because of the conflict of interest, the outcome will be a coordination failure. 54 MICROECONOMICS - DRAFT The Disagreement Game is similar to the Assurance Game in that: • There are two Nash equilibria • Both players do better if they coordinate (that is, speak the same language at one or the other of these equilbria) The Disagreement game differs from the Assurance Game because: • Each player in the Disagreement Game prefers one of the Nash equilibria while the second player prefers the other, while both prefer the Paretosuperior Nash equilibrium in the Assurance Game, so as a result • the players in the Disagreement Game have a conflict of interest concerning which equilibrium gets selected. Of course the English speaker would prefer to communicate in her home language and the Swahili-speaker feels the same way, but they would do much worse if they did not have any common language at all and if they failed to coordinate. Disagreement Games highlight how there can be social interactions with multiple Nash equilibria, each of which is Pareto-efficient, but there may be no ’middle ground’ to coordinate on and as a result conflict over who gets to benefit the most is unavoidable. Both players in the Disagreement Game would both be worse off out of equilibrium than at one of the Nash equilibria in the game. They have a common interest in coordinating somehow as opposed to not coordinating; but their interests conflict in how they coordinate. Checkpoint 1.14: Language Game Label the outcomes of the Language Game as in Figure ??, plot them using axes with the players’ payoffs, and determine which outcomes are Nash equilibria and which are Pareto-efficient. 1.13 Why history (sometimes) matters As we have seen from Disagreement Games and Assurance Games, strategic complementarities in games may give rise to more than one Nash equilibrium. When this is the case we cannot say which Nash equilibrium is our prediction of how the game will be played. The best the Nash equilibrium concept could do is to say that the outcome of the game is likely to be one of the (perhaps many) Nash equilibria. We need more information to make a prediction. Think about the Palanpur game, and imagine that all you know is the payoff matrix (not how the farmers played the game in recent years). Though you would be on solid ground H I S TO RY One of the first game theoretic studies of coordination problems – by David Lewis – was concerned with how we coordinate on a common language.8 S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS 55 predicting that it is likely that you’d see both farmers planting either early or late, you would not have much confidence in which it would be. But now suppose you were told that last harvest they planted late. Then, unless they had discovered some way to coordinate a switch to planting early, you would be correct when you predicted that they would both be planting late this year too. When history matters in this sense, we say that outcomes may be pathdependent. When the outcome of a game is path-dependent, without PATH DEPENDENCE A process is path dependent if the most likely state of something this period – the fraction of farmers planting early or late in the example – depends on its state in recent periods. knowing the recent history of a social interaction we cannot predict which equilibrium will occur. So, quite different outcomes – poverty or affluence, for example – are possible for two interacting groups of participants with identical preferences, technologies, and resources but with different histories. This is how "history matters." The Palanpur payoff matrix describes a poverty trap. A poverty trap occurs when identical people in identical settings may experience either an adequate living standard or poverty, depending only on chance events of their histories, for example were their parents rich or poor, or were they citizens of Norway or Nigeria. The possibility of poverty traps alerts us to the fact that people may be rich or poor not because of anything distinctive about their skills, hard work or other personal attributes, but because of the situation they find themselves in. Poverty may be inherited as it is in Palanpur not by anything that parents pass on to children but instead by the inheritance of a common history. The same is true about other aspects of how we interact in society, for example in the ways our lives may be highly segregated in interacting with people who differ in the groups with which they are identified, whether that be ethnicity, or religion , or even loyalty in sports teams. Checkpoint 1.15: Drain the meadow: Name that game Write down a payoff matrix for Hume’s "drain the meadow" game, with the two actions open to farmers Adams and Brown being "drain" and "do not drain", and assuming that the value of the drained meadow (to each farmer) is 5, the value of the undrained meadow is 3, and if the two farmers jointly work on the draining it costs them 1 each, while if a single farmer does the draining alone it costs him 3. What kind of game is this? Explain how it might be solved if there were just two farmers, and why with many farmers (as Hume wrote) it would be " difficult and indeed impossible" for them to agree on a common course of action and avoid in a coordination failure. P OVERTY T RAP A poverty trap occurs when identical people in identical settings may experience either an adequate living standard or poverty, depending only on chance events of their histories. Poverty in this case is a result of a person’s circumstances. 56 MICROECONOMICS 11 - DRAFT 12 1 10 2 9 3 8 4 7 6 5 (a) The circle as a clock 1.14 (b) The baseline Application: Segregation as a Nash Equilibrium among people who prefer integration Segregated communities – whether on grounds of ethnicity, race, religion, or class – are often the basis of inter-group prejudice and hostility and the systematic denial of equal dignity to all citizens. Segregation often results from deliberate policies of discrimination by governments, lenders, and citizens, as illustrated by the apartheid system of enforced racial separation in South Africa that persisted until 1994 and legally mandated housing segregation in (c) The citizens’ ideal integrated neighborhood Figure 1.16: The neighborhood and the citizen’s ideal integrated outcome Panel a is the "geography" of the neighborhood, showing that, for example, the citizen at position 2 on the circle has two immediate neighbors, the people at positions 1 and 3. Panel b shows that the person at position 2 is a Blue and her two immediate neighbors are both Greens. is just a starting point at which the neighborhood is as integrated as possible in the sense that the two immediate neighbors of each citizen are of the other type. Panel c shows the distribution of types across locations that the citizens prefer: each citizen has one immediate neighbor of each type. the U.S. – the so-called racial covenants the were finally outlawed in 1968.9 But segregation can also result from the uncoordinated decisions of people who would actually prefer to live in integrated communities. This counter-intuitive result illustrates the use of the Nash equilibrium concept, underlining the lesson already learned from the interaction among the Palanpur farmers: There may be more than one Nash equilibrium – one Pareto superior to the other – and a society can find it it difficult to escape the inferior equilibrium. The example of segregation is also a reminder – like the case of the over-fishing Nash equilibria– that the fact that an outcome is a Nash F AC T C H E C K In Seattle, Washington, what are termed “racially restrictive covenants” covering more than 20,000 homes prohibit sale or rental to particular groups. One stipulated that, “No person or persons of Asiatic, African or Negro blood, lineage, or extraction shall be permitted to occupy a portion of said property.”10 Racially restrictive covenants have been illegal since 1968 in the U.S. and are unenforceable. equilibrium does not mean that it is something that the players would choose, if they could coordinate and decide jointly on the outcome. The set-up of the model Here is a model. There are two types of people, Greens and Blues, and they live in homes arrayed around a circle representing a neighborhood. The homes are identical except that they may differ in the types of the immediate neighbors. The neighborhood is the circle as a whole. A household’s immediate neighborhood is made up of the two households on either side of it. Figure 1.17: Thomas Schelling (1921-2016) was an American economist who won the Nobel Prize in economics in 2005 for his contribution to our understanding of conflict and cooperation. The model we propose here is based on his work.11 S O C I E T Y : C O O R D I N AT I O N P R O B L E M S A B C (a) Household B has one neighbor of each type, their ideal situation, is not dis-satisfied, and will play Do Nothing A B C (b) Household B has two neighbors of their own type, is not dis-satisfied, and will play Do Nothing If a citizen would like to live at some other location around the circle, they can switch with some other person currently occupying that position, as long as the other person is willing. The homes just change occupants with no money changing hands. We would like to know what the neighborhood will look like when all the switching that people can do has been carried out, so that the neighborhood’s composition stops changing. The distribution of types among the houses around the circle when no citizen can benefit by moving is a Nash equilibrium. Greens and Blues have identical preferences about the type of their two immediate neighbors only. All people in the neighborhood would prefer to have one neighbor of each type, as is shown in Figure 1.18. But they are “satisfied” as long as they either have an immediate neighbor of each type or if both are of their own type. People are “dissatisfied” if both immediate neighbors are of the other type. An ideal neighborhood, then is shown above in Figure 1.16 c: Each person has one neighbor of each type. People have two strategies: “Do Nothing” or “Signal Dissatisfaction.” Signalling dissatisfaction means being willing to switch positions with another person – anywhere in the neighborhood – who has also “signalled dissatisfaction.” People are willing to switch only if they prefer the new location to their old location. For this reason people of either type will never switch with a person of the same type. This is because, for example, if a Green is dissatisfied with her current location and would like to move, all other Greens would be equally dissatisfied were they to take her position, so no other Green would agree to a switch. So all of the switches will be with different types: a green will switch with a blue, but a blue will never switch with a blue and a green will never switch with a green. This means that switches will change two things: • the switcher’s new immediate neighborhood: those on either side now experience having a neighbor of a new type given the switcher’s arrival and the previously dissatisfied person’s departure • the switcher’s old immediate neighborhood: those who were previously on either side of the switcher have a neighbor of a new type given the arrival of the person with whom the previously dissatisfied person switched. & ECONOMIC INSTITUTIONS A B 57 C (c) Household B has two neighbors of the other type, is dissatisfied, and will play Signal Dissatisfied Figure 1.18: The preferences of a household depending on the kinds of neighbors that surrounds it. Household B will either be satisfied or dissatisfied depending on the the types of neighbors they have. B’s choice to play Signal Dissatisfaction or Do Nothing therefore depends on the composition of their immediate neighborhood. 58 MICROECONOMICS - DRAFT A segregated Nash equilibrium We begin with 6 Greens and 6 Blues occupying alternating positions in the 12 “houses” at the locations on the circle that are numbered as if from time on a clock (so, 12 is the top). The twelve homes on the circle are "the neighborhood." We call the assignment of different types to the the homes around the circle in Figures 1.16 b and c: an allocation. An allocation in this game is an assignment of homes to the types at a given stage of the game. The allocation before the game starts is the initial allocation. The allocation after the game ends is the final allocation. The game proceeds as follows. At each step, each of the 12 people plays either Do Nothing, or Signal Dissatisfaction. Their choice of a strategy is known to all other players. Then, one of the twelve citizens is randomly selected and given the opportunity to make a switch if she can find another person willing who has also signalled dissatisfaction and is willing to make a switch. At step 1, for example, we might ask the Green at position 10 o’clock if she would like to switch. She would, because both of her neighbors are Blues. Whether she is able to make a switch depends on whether there are others who have chosen the strategy Signal Dissatisfaction. Because everyone else is also dissatisfied, she has many choices. Suppose she switches with her friend and immediate neighbor, the Blue at position 11, shown by the colors of position 10 and 11 changing from Panel a Start to Panel b Step 1. The two people are still friends and neighbors, but each now also has the same type of neighbors on the other side. Suppose, next, that it is the Blue at position 7 who is picked to stay or switch. If he plays "Signal Dissatisfaction," he could switch with his friend and immediate neighbor at position 6. We continue this process until either no one is dissatisfied, or if someone is dissatisfied, there are no others playing the strategy Signal Dissatisfaction with whom a switch is possible. This process could continue as shown in the figure, resulting at the end of 5 steps in the completely segregated neighborhood shown in Figure 1.19. Notice that over the course of the game, a particular home may change hands more than once. The home at position 7 for example started off occupied by a Blue who switched with a Green in Step 2, who then switched with a Blue in Step 4. At step 6 (not shown), each of the 12 would choose the strategy Do Nothing, because 8 of them have same type as neighbors only and the other four have one neighbor of each type. So no one is dissatisfied. As a result there we observe no further moves: the allocation is stationary (meaning unchanging). It is a Nash equilibrium. We could expand the strategies available to the players to allow those with S O C I E T Y : C O O R D I N AT I O N P R O B L E M S & ECONOMIC INSTITUTIONS (a) Start (b) Step 1 (c) Step 2 (d) Step 3 (e) Step 4 (f) Step 5: Complete Segregation both neighbors of their own type to signal dissatisfaction, and if a willing other citizen could be found, they could switch with this person so as to have one neighbor of each type. This will not disrupt the completely segregated neighborhood as long as people prefer to have both neighbors of their own type to having both neighbors of the other type. Avoiding outcomes that nobody prefers The conclusion is not that complete segregation will necessarily be the result. This is true for two reasons. • There is also a Nash equilibrium that is integrated rather than segregated. In Figure 1.16 c, the allocation has each person’s immediate neighborhood composed of both types. You can confirm that, like the completely segregated allocation, this integrated allocation is also a Nash equilibrium: every citizen has their ideal immediate neighborhood, so no citizen is dissatisfied and each are best responding with Do Nothing. This allocation could have 59 Figure 1.19: From integration to a segregated Nash equilibrium. The figure shows one of many possible progressions from an integrated non equilibrium situation to an entirely segregated Nash equilibrium. Panel a shows the starting point from the previous figure. In step 1 the Green at position 10 and the Blue at position 11 switch positions, shown by the double heded arrow, and resulting in the neighborhood shown in Panel b. The remaining panels show the successive steps to the final fully segregated Nash equilibrium. 60 MICROECONOMICS - DRAFT come about by the same rules of the game that resulted in complete segregation. This is an example of implementing a desirable allocation within given set of rules of the game • The citizens could play the game cooperatively rather than non-cooperatively. If the citizens had realized that playing the game non-cooperatively could lead them to a complete segregation outcome that nobody wanted, they could have acted cooperatively – that is jointly agreed – to implement their ideal allocation. This is an example of implementing a desirable allocation by changing the rules of the game: agreeing to act jointly was not an available strategy in the non-cooperative variant of the game above. The outcome in the segregation model shares three features with a game from what would appear to be a very different situation: Planting in Palanpur. • A Pareto inferior Nash equilibrium: There is a Nash equilibrium – planting late and a segregated community – in which everyone is worse off than they could be at some other allocation. • A path-dependent outcome: History matters because an outcome that is preferred by all participants is also a Nash equilibrium, so if the preferred outcome were to occur, it could persist. • A change in the rules of the game can avoid the inferior outcome: By coordinating their actions – changing the interaction to a cooperative game – they could escape the Pareto inferior outcome In these three respects the two interactions – when to plant and where to live – are not unique or even unusual in these three respects. Checkpoint 1.16: Segregation as a Nash equilibrium a. Show that the segregated neighborhood in Figure 3 is a Nash equilibrium. b. Show that the ‘ideal neighborhood’ in Figure 1 is also a Nash equilibrium. c. Show that in Figure 1.19 if Step 3 had been different the equilibrium allocation could have been the citizens’ ideal integrated allocation. Which Step 3 switch would have accomplished this result? d. Suppose that the game was changed slightly so that a dissatisfied person knows only the “dissatisfaction status” of her immediate neighbors. Show that, starting with the alternating types status quo (Panel a Start, in Figure 1.19 ) that the neighborhood would then evolve to the ideal distribution. e. Suppose that in the fully segregated neighborhood case citizens decided to have a binding referendum to implement the ideal neighborhood (requiring whatever moves are necessary to bring that about), but because the question of where you live is a sensitive one, they adopt the rule that unanimous approval of the referendum is required for it to be implemented. Would it be implemented? S O C I E T Y : C O O R D I N AT I O N P R O B L E M S 1.15 How institutions can address coordination problems Game theory has given us a catalogue of coordination problems: Prisoners’ Dilemmas, Invisible Hand Games, Assurance Games, and Disagreement Games (there are many more!). Knowing how the structure of these games differ will help to diagnose the nature of a coordination problem and to devise policies and constitutions – changes in the rules of the game – to avoid a coordination failure. This is an example of using the concept of equilibrium to understand how to change an undesirable outcome. The idea is simple: a change in the rules of the game can eliminate an undesirable Nash equilibrium, so that it is no longer our prediction of how the game will be played. Instead it may be possible to make some preferable strategy profile a Nash equilibrium which then could be the outcome of the game. A common approach to averting coordination failures is in a Prisoners’ Dilemma to devise policies or institutions that transform the payoff matrix so that the game is no longer Prisoners’ Dilemma. There are two possibilities to consider: • Change the Prisoners’ Dilemma to an Assurance Game • Change the Prisoners’ Dilemma to an Invisible Hand Game Changing the Prisoners’ Dilemma game into an Assurance Game means making the mutual cooperate outcome a Nash equilibrium (it was not in the Prisoners’ Dilemma) even if mutual defect remains a Nash equilibrium. The second options is a more ambitious policy objective: converting a social interaction from a Prisoners’ Dilemma to an Invisible Hand Game. To see how this might work, remember that the coordination failure that results in the Prisoners’ Dilemma is a consequence of the fact that in that game players can take actions that inflict costs on others – negative external effects – that are not part of their thinking when they decide what to do. To see that internalizing these external effects can address the coordination failure, we examine the implementation of a liability rule in the Fishermen’s Dilemma. Tort is a branch of law dealing with damages inflicted by one person on another (or another’s property). Among other things, tort law establishes the responsibility – called liability – of the person inflicting the damages to compensate the harmed individual. The requirement to compensate the harmed individual internalizes the external effect. How would a liability system work in the Fishermen’s Dilemma? Look again at Figure 1.12, the general form of the Prisoners’ Dilemma. Suppose Alfredo and Bob decided to jointly adopt “Cooperate” (fish 10 hours) as an agreement. In their agreement, they also choose to adopt a liability rule requiring compen- & ECONOMIC INSTITUTIONS 61 62 MICROECONOMICS - DRAFT Bob 3 2 12 Hours 3 2 10 Hours 2 2 (a) Liability rule with numbers (Defect) (Cooperate) 3 3 10 Hours (Cooperate) 12 Hours (Defect) Alfredo 10 Hours (Defect) (Cooperate) 12 Hours Alfredo 10 Hours (Cooperate) Bob 12 Hours (Defect) y − (w − x) w ● ● w w w y − (w − x) z z (b) General liability sation be paid to the other party if one’s actions result in lower payoffs than would have occurred had the agreement to cooperate (fish only 10 hours) been observed. With the liability rule the following will happen: • If Alfredo defects on Bob (plays Fish 12 hours), Alfredo initially gets y as before Figure 1.20: Fishermen’s Dilemma with a liability rule. Players can implement a desired outcome by Transforming Property Rights using a liability rule (the harm a player does to another player is deducted from their payoff). This payoff matrix is based on Figure 1.12 modified by the liability rule. Alfredo’s payoffs are listed first in the bottom-left corners and shaded blue. Alfredo’s best responses are shown by the solid point. Bob’s are listed second in the top-right corners and shaded red; his best responses are shown by the hollow circle. • But then Alfredo must compensate Bob for the costs his defection has inflicted, that is, Alfredo must pay compensation sufficient to give Bob a payoff of w (which is the payoff Bob would have obtained had Alfredo not violated the agreement) • If both players defect, they both gain z • But, then they must compensate the other player by a transfer of w z. We can use these changes to the payoffs to construct a transformed payoff matrix. The transformed payoff matrix for Alfredo’s and Bob’s payoffs is given by the entries in Figure 1.20. Did the improved property rights succeed? Because y w + x < w by the def- inition of a Prisoners’ Dilemma, Cooperate is now a best response to Cooperate and (Cooperate, Cooperate) is a Nash equilibrium. Cooperate is also a best response to Defect (because w > z), so Cooperate is the dominant strategy with the liability rule in force, and (Cooperate, Cooperate) is the dominant strategy equilibrium. The Prisoners’ Dilemma game has become an Invisible Hand game through the adoption of a new set of property rights. Redefining property rights – to take account of liability for damages – can implement a Pareto-efficient outcome by inducing each player to account for R E M I N D E R For a Prisoners’ dilemma we need y > w > z > x and x + y < 2w. It’s the second one that’s important here. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S how his actions effect the other player. By redefining property rights to include the liability of the damages (external effects) that one inflicts on others, we have transformed the game to an Invisible Hand game. M-Note 1.1: The mathematics of the liability rule Refer to Figure 1.20. For the original game to be a Prisoners’ Dilemma requires: • y > w > z > x, • x + y < 2w, • and has the Nash equilibrium (Defect, Defect) with payoffs (z, z). The following inequalities must help us to think through the logic of the liability rule: • With the transformed payoffs the Nash equilibrium must be (Cooperate, Cooperate) with payoffs (w, w). • The condition for x + y < 2w arises because we require that for w > y (w x). We can re-arrange the condition because w > y w + x, therefore, by adding w to both sides we get 2w > y + x as we said we need to assume. Be sure that you can identify the intermediate step here that we have skipped in the matrix: For the combination (12 Hours, 10 Hours) and (10 Hours, 12 Hours) the payoff to the player playing 10 hours is w because it was x + (w with x, but were then rewarded with the transfer for (w the agreement. x + (w x) (that is, the players started x) by the other player breaking x) = w! Notice too, that both of the players receiving z in the (12 Hours, 12 Hours) occurs because they both compensate the other with (w z (w z) + (w z) and z ) = z. Checkpoint 1.17: Limited Liability by the Numbers Use the model of the liability rule in Figure 1.20 to complete the following tasks. a. Re-draw the payoff table, but substitute in the values for x, y, w and z from Figure 1.10. Hint: The payoffs should only be 2s and 3s. b. Solve your new game using best response analysis (the circles and dots method) to find the Nash equilibrium of the game. What is it? Explain. c. Does either player have a strictly dominant strategy? Is there a dominant strategy equilibrium? Explain. 1.16 Game theory and Nash equilibrium: Importance and caveats We have started off this introduction to modern microeconomics with game theory. The reasons are that • Many important economic relationships – in labor markets, families, credit and financial markets, between citizens and governments, among neighbors, between nations seeking to address climate change, and many more – are strategic, and require the tools of game theory. & ECONOMIC INSTITUTIONS 63 64 MICROECONOMICS - DRAFT • Focusing on people as actors often in conflict with each other, but also sharing common interests, is essential to economics as a social science, and game theory allows us to do this. • How these interactions work out depends on the institutions that regulate them, and game theory allows us (even requires us) to be very specific about the varieties of possible rules of the game under which we now interact, and how we might change these rules for the better. For game theory the Nash equilibrium is a key economic idea and it provides a way to answer the question: what will be the outcome if each of the actors adopts a strategy that will not lead any other actors to change what they do? In many situations the Nash equilibrium among players who seek to maximize their own material payoffs provides a good prediction of what we observe in the real world. But not always. • Extreme individualism: Overlooked opportunities to coordinate: If the two fishermen were fishing 12 hours each and the details of the situation were such that they could just agree to fish 10 hours instead, then we would be mistaken to use the Nash equilibrium of this game to predict what the players would do. In this case the rules of the game have been inadequately described: there was a third strategy that was overlooked, namely agree with the other to fish 10 hours as long as the other agreed to the same. With the game modified in this manner, the Nash equilibrium would provide a good prediction: both Bob and Alfredo would choose "Agree" as long as they believed that the other would do the same and that the agreement would be enforced. • Self-interest: Overlooked payoffs not in the form of one’s own material gains: Even if for some reason they could not agree on a common course of action, if Alfredo and Bob were brothers and each cared about how much fish the other had to feed his family, then the two would not define their payoffs in the game simply as amount of fish they each caught individually, but would include the consequences of their actions on the outcome for the other fisherman. So we would have to rethink the payoff matrix. In this case it is the payoff matrix that was mistaken. It no longer describes what the players are trying to maximize, that is the incentives that are shaping their behavior. • Selection among many equilibria: As we have seen in the Planting in Palanpur Game, there may be more than one Nash equilibrium, so the prediction that the result of the interaction will be a Nash equilibrium is insufficient. We need to know more about the situation – such as the recent history of the people involved – to make a prediction. E X A M P L E When people can bind themselves to common agreements or when they have values other than maximizing their own material gains, we need to modify the game as we will see later in this chapter, and in Chapter 2. S O C I E T Y : C O O R D I N AT I O N P R O B L E M S • Dynamics: Sometimes we are more interested in the process of change itself than in the stationary end point of this process. Whether studying how the people of Palanpur might make a transition from poverty to affluence, or how the polar sea ice might collapse due to climate change we are often interested in how the economy works when it is not in equilibrium. 1.17 Application: Cooperation and conflict in practice If all that is needed to address a coordination failure is to require that people pay the costs that their actions impose on others then why are coordination failures so common? Over-exploitation of fisheries is an international problem that humans as a world community have failed to solve. Many over-exploited fisheries will not recover for a long time. But local communities and groups of fishermen have found ways to combat over-fishing and we can learn many lessons from what they do. Many groups – from farmers to fishermen – face equivalent problems worldwide. These outcomes provide a concrete motivation to study the Prisoners’ Dilemma Game and other coordination problems. What we learn from these games is that an effective liability rule requires two things: • The injured party or the courts have to have verifiable information (that is information sufficient to enforce the liability aspect of the property right) and • There has to be a court or some other body willing to and capable of enforcing the contract. When we turn from game theory to the study of real fishing communities we find that both conditions are unlikely to be met, which is why the overexploitation of fisheries continues in many cases. • Limited information. The lack of verifiable information is common in social interactions and this limits the policies that governments or private actors can design in response to the persistence of Pareto-inefficient Nash equilibria. • Conflicts of interest. Governments may not have the capacity or the will to enforce the necessary policies especially in cases where doing this would impose costs on a powerful group. An example is the failure of many countries to address the problem of climate change, which is in part the result of the fossil fuel companies’ opposition to putting a sufficiently high price on carbon emissions. Fishing communities, of course, are not acting out a tragic script, as were the herders in Hardin’s tale about the tragedy of the commons. They are not & ECONOMIC INSTITUTIONS 65 66 MICROECONOMICS - DRAFT prisoners of the dilemma they face. Real fishermen are resourceful and seek solutions to the problem of over fishing. • Lobstermen in the U.S. state of Maine limit how many lobster they catch using highly local restrictions on who can set traps where (the state government provides the legal framework for this).12 • Turkish fishermen allocate fishing spots by lottery and then rotate the spots so the distribution is fair.13 • The fishing community of Kayar in Senegal adopted the rule that only one trip to the fishing grounds per day is permitted (a bit like Alfredo and Bob limiting their hours of fishing) and appointed a committee to check that the rule was being observed. They also limited the number of boxes of fish that could be offloaded by a single canoe.14 • Shrimp fishermen in Toyama Bay, Japan have a rule that they offload their daily catch at the same time and place, so that the size of each boat’s catch would no longer be asymmetric information.15 These rules and practices based on small local fishing communities are often disrupted by the entry of other groups who are not bound by the local rules. Conflicts of interest within the local community also sometimes limit the effectiveness of attempts to limit the catch. One reason is that restrictions on fishing are often supported as a way to raise the wholesale price of fish and hence the incomes of fishing families. But fish sellers – who buy the fish wholesale at the port and then sell to local consumers – would profit if they could pay less. The rules regulating access to fishing that we see in existence are a small selection from a much larger set of rules that people have tried out at some point. What we see are the institutions that have succeeded well enough to allow the communities using them to persist and not to abandon their rules. The persistence of such rules does not require the rule to implement a Paretoefficient outcome, it only requires that the rule be reproduced over time by people adhering to it. By this reasoning, even if the rules of the game do not implement Pareto efficient outcomes, we might expect a fishing community that has hit on a way of sustaining cooperation in the long run to do better in competition with groups that over-fish, and that successful groups may be copied by other groups. Checkpoint 1.18: Institutions and Palanpur Supposing that the only voters involved in approving the Palanpur village council’s decision to require planting early were themselves farmers, explain why they would unanimously support the measure. What would happen if after implementing the law requiring early planting one year, the next year the law was S O C I E T Y : C O O R D I N AT I O N P R O B L E M S taken away? 1.18 Conclusion The classical institutional challenge which we stated originally was “How can social interactions be structured so that people are free to choose their own actions while avoiding outcomes that are worse for everyone than they could be?” With the terms you have learned this can now be re-phrased “How can social interactions be structured so to avoid Pareto-inefficient Nash equilibria resulting from people’s free choice of their own actions?” The Fishermen’s Dilemma is an example of a challenging coordination problem because an inefficient outcome is the unique Nash equilibrium. To study a game and its likely outcomes and also how to improve these outcomes we have proceeded in three steps: • First, use the Nash equilibrium concept to identify one or more likely outcomes of the game • Second, use Pareto comparisons to identify outcomes that are "worse for everyone," and • Third, devise changes to the relevant institutions – the rules of the game – or that would shift the population to a superior Nash equilibrium either preexisting (as in the case of Planting in Palanpur, or the segregation case) or novel (as with the transformed Prisoners Dilemma Game). We have illustrated the third step by a legal remedy, the introduction of tort liabilities for damages in the Prisoners’ Dilemma Game so as to internalize the external effects accounting for the coordination failure. But an adequate analysis of coordination problems and their possible mitigation must do two things that we have not yet taken up: • Evaluate outcomes according to their fairness and other ethical criteria and • illuminate the far broader range of how markets, governments, private associations like firms or neighborhoods and other bodies might jointly accomplish this task. We take up this challenge in Chapters 2, 4 and 5. Making connections Institutions and the rules of the game: To predict or explain the outcome of a social interaction, it is essential to know the “rules of the game” that determine who knows what and when, who gets to do what and when and as a result who gets what. & ECONOMIC INSTITUTIONS 67 68 MICROECONOMICS - DRAFT Equilibrium: Equilibrium describes an outcome that will persist until some aspect of the situation changes as a result of externally caused changes. A Nash equilibrium is a special kind of equilibrium widely used in economics. External Effects: People often take actions without considering the effects of these actions on others. The resulting external effects – positive and negative – pervade social interactions. Pareto inefficiency of Nash equilibria: A common result is that the outcomes of social interactions (the Nash equilibria of the games) are Paretoinefficient meaning that opportunities for mutual gains remain unrealized. Rents: When players interact they face opportunities for mutual benefit, or common interest. But this creates opportunity for rents and for a conflict over how the benefits resulting from the interaction will be distributed. Policy and changes in the rules of the game: Improving property rights (such as making people legally responsible for the harms to inflict on others) can lead people to internalize external effects. Other institutions that would facilitate people making decisions to act jointly can also provide solutions to coordination problems. Policy may result in a shift to a Pareto-efficient equilibrium. Positive feedbacks, increasing returns, and strategic complementarity: Often players strategies are strategic complements due to positive feedback and increasing returns. As a result, in some social interactions there may may be many equilibria as in the Assurance Game and the Disagreement Game. Important Ideas institutional challenge fallback Pareto-superior/inferior coordination problem next best alternative Pareto efficient player best response (weak/strong) (economic) rent strategy dominance (weak/strict) payoff Dominant strategy equilibrium institution Nash equilibrium interdependence positive external effect negative external effect Prisoners’ Dilemma (non-)cooperative games optimization Tragedy of the commons Invisible Hand Game Assurance Game Disagreement Game Increasing returns Strategic complement/substitute Path dependence Poverty trap Liability rule positive feedback S O C I E T Y : C O O R D I N AT I O N P R O B L E M S Mathematical Notation Notation Definition p a player’s payoff Note on superscripts: A, B: individuals. Discussion questions See supplementary materials. Problems See supplementary materials. & ECONOMIC INSTITUTIONS 69 2 People: Self-interest and Social Preferences DOING ECONOMICS How selfish soever man may be supposed, there are evidently some principles in his nature, which interest him in the fortune of others, and render their happiness necessary to him, though he derives nothing from it, except the pleasure of seeing it. Adam Smith, The Theory of Moral Sentiments (1759) Chapter 1. Chicago, a city famous for its pizza, sports, jazz, and its skyline, built its fortune on the farming of the state of Illinois. Today Illinois farmers use high tech machinery and advanced business plans, some cultivating a thousand acres of land or more. But many of the farmers don’t own the land they cultivate; they rent land and work it as a sharecropper. Sharecroppers are farmers – "tenants" – who pay the owners of land a share of the total harvest that they cultivate. In the mid-1990’s, over half of the contracts between farmers and owners were sharecropping agreements, and in Northern Illinois 95 percent of these contracts stipulated a fifty-fifty division of the crop between the owner and the sharecropper. An equal split of the crop means that a tenant on fertile land will have higher income for the same amount of effort and other inputs.1 Because a tenant on fertile land will reap a larger harvest than a tenant on poor quality land, the fifty-fifty sharecropping contract presents us with a puzzle. Here’s the puzzle: if half of the crop on poor quality land is sufficient to attract tenants, why should the owners of good quality land give up half of the crop to their tenants? Those tenants must be earning more than what was necessary to get them work the owner’s land. So, why don’t the owners of the better land propose a tenant’s share less than half, giving the tenants just enough so that This chapter will enable you to: • Understand that people make decisions based the actions open to them (constraints), which of these possible actions they believe they must take (beliefs) to bring about the outcomes they most prefer (preferences). • Use this approach to analyse economic situations involving risky outcomes, bargaining, and contributing to the public good. • Analyse sequential games and games with multiple Nash equilibria, showing how being the first mover in these games may confer advantages on a player. • Explain how the experimental method along with this "preferences, beliefs, and constraints approach" is used to study economic behavior. • See how changes in the rules of the game can result in better outcomes for all. • Describe the experimental and other evidence that people’s preferences go beyond self-interest and include generosity, reciprocity, fairness and hostility toward others . • Understand that these other-regarding preferences are as much part of what we consider to be rational action as is self interest. • Give examples of the importance of social norms and culture for decisionmaking and economic policy-making. they are willing to farm the land? We would expect owners to insist on lower tenant’s shares to sharecroppers on higher quality land and offer higher shares to sharecroppers on low quality Figure 2.1: Farming in Illinois is big business. 72 MICROECONOMICS - DRAFT land. Because land varies in quality by small gradations, this would result in a pattern of sharecropping contracts with a range of shares depending the land quality. But they is not what we see. Almost all of the contracts are fifty-fifty. Illinois sharecropping contracts allow the sharecroppers on good land to receive income attributable to the superior land quality, income the owners would otherwise have received if the owners had insisted on a lower tenant’ s share on the high quality land. The fifty-fifty split effectively transfers millions of dollars annually from owners to sharecroppers simply because of the fifty-fifty division. This is not a peculiarity of Illinois. Fifty-fifty is the norm in sharecropping around the world. Rice cultivation in West Bengal, India during the 1970s provides another example. There, poor illiterate farmers in villages isolated by impassable roads for much of the year and lacking electronic communication eked out a bare living on plots that average just two acres rather than the thousand-acre plots farmed in Illinois. The Indian farmers shared one similarity with farmers in Illinois: the division between sharecroppers and owners was fifty-fifty in over two-thirds of the contracts.2 Why was the contract the same in these distant parts of the world? The short answer is that where most contracts are fifty-fifty, that particular division is a social norm, something people feel they are morally obliged to respect. The fact that around the world land owners respect a social norm that overrides their material self-interest tells us that many people are committed to acting fairly, being treated fairly and conforming to ethical standards of appropriate behavior. But the sharecroppers in Illinois and West Bengal, like farmers everywhere, are also trying to make a decent living, or even to become affluent. They are not simply following social norms. They carefully weigh alternative methods of cultivating their crops at the least possible cost and marketing their harvest at the highest possible price. 2.1 Preferences, beliefs and constraints Understanding economic behavior requires a model that takes account of what people care about ( for example, the farmers’ incomes, and also their desire to uphold social norms) and how from the actions they are able to undertake, they adopt those that are they think will bring about desired results. We will develop a model of economic behavior based on: • constraints: the feasible set of actions, meaning actions that are open to us, H I S TO RY In 1848 the British philosophereconomist John Stuart Mill noted the striking global pattern of equal division in crop sharing, calling it “the custom of the country” and “the universal rule.”3 PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES Beliefs Preferences (which actions result (evaluation of in which feasible outcomes) feasible outcomes) 73 Figure 2.2: Preferences, beliefs and constraints. The actor may choose from a set of feasible actions (the constraint set on the left). Combining that set with her beliefs about the outcome produced by each of the actions in the the constraint set, she then has a set of outcomes that she believes are feasible, depending on her choice of an action. From all of these outcomes in the set believed to be feasible, she identifies the one that is ranked highest according to her preferences and then takes the action that she believes will bring about this outcome. Set of Constraints (Feasible set of actions) outcomes believed to Choice of an action be feasible • beliefs: our understanding of the outcomes that will result from the actions that are open to us,and • preferences: our evaluation of the outcomes that we believe will result from the actions we take. This is called the preferences, beliefs and constraints approach. The relationship between these three elements of the preferences, beliefs, and constraints approach is described below and is shown in Figure 2.2. Game theory, which you have already studied, is an important example of the preferences, beliefs and constraints approach. Constraints: limits on action From a long list of things she might consider doing, constraints define a more limited possible set of actions, namely the shorter list of all of those so called feasible actions she can carry out. In the game theory introduced in the previous chapter the constraint was the set of possible actions, that is, a list such as "Fish 10 hours", "Fish 12 hours" or "Plant early", "Plant late". Constraints may be imposed by personal limitations, by laws of nature, or by the force of law. A constraint can also reflect a decision by the actor to eliminate some action from the feasible set of actions on moral grounds, irrespective of the payoffs. Examples are keeping promises, or not committing murder. In Table 2.1 we give examples of how the preferences beliefs and constraints approach can be applied. The list of feasible actions set by constraints need not be just a list of particular actions, like drive or take the bus. When marketing their output (first row of E X A M P L E The preferences, beliefs and constraints approach is sometimes called rational choice theory or the rational actor model, but we prefer the more specific label that we use here as it identifies the the three important elements making it up. P REFERENCES , BELIEFS , AND CONSTRAINTS APPROACH According to this approach, from the feasible set (which includes all of actions open to the person given by the economic, physical or other constraints she faces) a person chooses the action that she most values as given by preferences, in light of given her beliefs about the actions that will bring about the outcome. 74 MICROECONOMICS - DRAFT Constraints Actor (feasible set of actions) Beliefs (information Feasible Outcomes about which actions (states that could will result in the result from preferred state) the actions) The demand curve Firm owner High or low prices (how quantity depends on price) Urban resident Ordering a meal Drive or How many others take the bus will drive The menu; your budget Simple: just order the best you can pay for Various levels of profits Travel time Meal quality, money left over the table), the owners of a firm, for example, can set any price they like (anywhere from 0.00 by penny increments up to some very high number). Wealth, the availability of credit, and the prices of goods impose constraints Preferences (ranking of all outcomes) Maximize profits Minimize travel time Maximize utility Table 2.1: Applications of the preferences, beliefs, and constraints framework. Real choice situations are typically not as simple as Figure 2.2. The urban resident, for example, may care both about travel time to work and his carbon footprint. on an actor’s consumption. The institution of private property also imposes limits: it means that theft is not an option for increasing your consumption. Given private property and in the absence of gifts or other transfers from a government, the total amount of goods and services you can consume is limited by wealth and how much you can borrow. So when we study someone’s consumption, their budget constraint is a critical factor as people have a certain budget determined by wealth, access to credit, and prices all limiting how much they can buy. Beliefs: Translating actions into outcomes Beliefs are a person’s understanding of the outcomes that her actions will bring about. In many cases what I must do to get the outcome that I prefer depends on what other people do. I would like to spend the evening with friends, but where I should go to make it happen depends on where I think my friends will go. Given that I cannot communicate with my friends (the batteries to their phones have run out), my action (where I will go) will therefore depend on my belief about where I will find my friends. In Table 2.1 the owners of firms are not constrained to set any particular price, but if they want to translate their choice of a price into what they care about – profits – they must form an opinion about the number of units they will be able to sell at each price. This is the demand curve, and it expresses the owners’ beliefs about the relationship between their action (the price) and an outcome (how many goods they will sell). In the game theoretic approach of Chapter 1 beliefs were expressed in the B ELIEFS A persons understanding of the relationship between her actions and the outcomes that will occur as a result of her actions are her beliefs. Beliefs are thus a causal mapping from the actions one can take to the outcomes that will occur. Where the outcomes of actions are not known with certainty, beliefs include probabilities of results occurring. E X A M P L E The word "belief" is often used to refer to spiritual matters ("religious beliefs"); but in game theory a belief is a statement about how the world works, namely what action is required to bring about some particular outcome. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 75 solution concept, that is a description of how the game would be played. The Nash equilibrium as a solution concept, for example, is based on the idea that players best respond to the play of the other players. This is the basis of players’ beliefs about how their choice of an action will translate into an outcome of the game. Preferences: Reasons for preferring one outcome over another Preferences are evaluations of outcomes that provide motives for actions. A person’s preferences are the reason why she takes the action that that she believes will bring about the outcome that is better than or at least as good as the others. In Chapter 1, preferences were represented by the payoffs in games that people played. For each player, a strategy profile was associated with a number – her payoff – and players chose actions that they believed would result in the strategy profile with their most preferred (highest payoff) outcome. In many games preferences are represented by money payoffs. But more E X A M P L E While most widely used in economics, the preferences, beliefs, and constraints approach is also used in political science, for example, to understand the strategies followed by elected officials seeking to maximize their chances of reelection, in law to design criminal or civil penalties to effectively deter illegal activity, and even in biology to study the evolution of genes, modeled as if they are "trying to" increase their numbers. P REFERENCES Preferences are evaluations of outcomes that provide motives for taking one course of action over another. broadly, preferences represent the favorable (positive) or unfavorable (negative) feelings a person has about an outcome that lead them to try to make an outcome happen (high payoff) or that lead them to try to avoid an outcome (low payoff). Preferences include: • tastes (food likes and dislikes, for example), • habits (or even addictions), • emotions (such as anger and disgust) often associated with visceral reactions (such as nausea or an elevated heart rate), • social norms (for example, those that induce people to prefer to be honest or fair), and • psychological tendencies (for aggression, extroversion, and the like). The difference between preferences and beliefs is simple. A preference says: I like the outcome X more than the outcome Y. A belief says: I believe I can get X to happen if I do some action Q. Self-regarding and other-regarding preferences A feature of the preferences, beliefs, and constraints approach is that it allows us to model choices based on the entire range of preferences whether they be entirely self-regarding, caring for others (wishing them well or wishing to harm them), or reflecting religious commitments. A key distinction about our preferences is whether in evaluating the results that we believe our actions will bring about (the right hand part of Figure 2.2) we think about the results that we ourselves experience only, or do we OTHER - REGARDING PREFERENCES A person with other-regarding preferences when evaluating the outcomes of her actions takes into account the effects of her actions on the outcomes experienced by others as well as the outcomes she will experience. 76 MICROECONOMICS - DRAFT also consider the results that are experienced by others. This gives us two categories of preferences: • If we think only about the results experienced by ourselves, we have selfregarding preferences • If we also think about the results experienced by others, then we have other-regarding preferences. Is the the same thing as "selfish" and "unselfish" preferences? No. Abraham Lincoln is said to have remarked: “When I do good, I feel good. When I do bad, I feel bad. That is my religion.” Does this mean that Lincoln’s "good" acts were in fact self-regarding because they made him feel "good?" SELF - REGARDING PREFERENCES When choosing an action, a self-regarding actor considers only the effect of her actions on the outcomes experienced by the actor, not outcomes experienced by others. A self regarding actor ignores the external effects of her actions on others. That does not follow. He had other-regarding preferences leading him to act differently than if he cared only about the outcomes that he personally experienced. In the preferences, beliefs and constraints model all actions are motivated by preferences, so doing a preferred thing cannot be termed "selfish" without making all behavior selfish by definition. That is why we use the term self-regarding rather than "self-interested" or "selfish." For example, if you (like Lincoln) enjoy helping others, and you act on these preferences, does this mean you are selfish (because, for example that’s what gives you a sense of leading a good life). No, it does not. You are acting on your preferences, but they are other-regarding because you enjoy trying to make the results that others experience be what they would want. Of course otherregarding preferences include feelings of altruism towards others, but they also include negative feelings about others, such as envy, spite, racism and homophobia. H I S TO RY In 1977 Amartya Sen wrote "Rational fools" in which he pointed out that the preferences beliefs and constraint approach ignores the importance of promises, what he called commitments. The reason is that the approach seeks to explain behavior entirely on the basis of the actor’s anticipation of what her actions will bring about in the future. Honoring a past commitment – not because she would otherwise feel guilty in the future, but because it is the right thing to do – cannot be modeled in the preferences beliefs and constraints approach. In sections 2.8, 2.9 and 2.12 we provide some evidence from experiments on the kinds of other-regarding preferences and how common they are across the world. "Rationality" The term rationality in economics means acting on the basis of: • Complete preferences This means, that for any pair of possible outcomes that a person’s actions may bring about, A and B, it is the case that the person prefers A to B or B to A or is indifferent between the two. Preferences are not complete if there is some other pair, say A and D for which none of the above three comparisons can be made: to the three statements “I prefer A to D," “I prefer D to A" and “I am indifferent between A and D" the person responds "none of the above." • Consistent preferences If an individual with consistent (also called transitive) preferences prefers a bundle of goods A to another bundle B, and bundle B to a third bundle, C, they also prefer A to C. R ATIONAL A rational person has complete and consistent (transitive) preferences and can therefore rank all of the outcomes that their actions may bring about (better, worse, equal) in a consistent fashion. C OMPLETE PREFERENCES Complete preferences specify for any pair of possible outcomes that a person’s actions may bring about, A and B, whether A is preferred to B, B is preferred to A or they are equivalent. Using the symbolic notation for preference: A B, or B A, or A ⇠ B. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES A person with complete preferences, which requires only that she can rank all pairs of outcomes, might nonetheless violate the consistency assumption. So she could prefer A to B, B to C, and C to A. All that matters for completeness is that she can rank each pair. 77 INDIFFERENCE When a person is indifferent between two outcomes, it is because those outcomes provide them the same payoffs, or the same expected payoffs. As a result, a person will not care which of the two (or more) outcomes they obtain between (or among) which they are indifferent. In the heading at the start of this section, we put quotation marks around rationality to underline the difference between how economists use the term and how it is generally used, that is to mean "based on reason." In everyday usage, the term "rational" often means something like the intelligent and perhaps even amoral pursuit of one’s own interests. But in economics, as you can see from the above definition, it means something entirely different. • Rationality does not say anything about what it is that the person values: A completely generous and ethical person is rational as long as her preferences are consistent and complete. • Rationality does not mean being intelligent or well informed: The beliefs that (along with preferences) determine the choices a person makes need not be true. Moreover, people with incomplete preferences would hardly be called "irrational" in the ordinary meaning of that term, meaning "not logical" or "unreasonable." Ask yourself if your preferences are complete for the following outcomes: express preference or indifference over which of your two dearest friends will be tortured to death. If you were to say "I cannot rank those two outcomes, nor am I indifferent between them" you would not be "rational" by the economic definition, but nobody would think your behavior was bizarre either. We might be more inclined to worry about the person who would be able to make such a ranking. Checkpoint 2.1: Why beliefs matter Considering the coordination problems studied in Chapter 1 a. Explain why in the Assurance Game representing planting in Palanpur why the action a farmer takes to bring about the preferred outcome depends on the farmer’s belief about what other farmers will do. b. In the same game explain why the farmer who believes most other farmers will plant late, will also plant late. c. Explain why Ben’s belief about what Aisha will do matters for how he will play in the Disagreement Game. d. Are there any games you have learned so far in which beliefs about what the other does did not affect the outcome of the game? C ONSISTENT ( OR TRANSITIVE ) PREF ERENCES Preferences are consistent (transitive) if whenever an individual prefers a bundle of goods A to another bundle B, and bundle B to a third bundle, C, they also prefer A to C. Using the symbol A B to mean "A is preferred to B" and the symbol ) to mean "implies", the condition for consistency can be written as: A B and B C ) A C. 78 MICROECONOMICS - DRAFT 2.2 Taking risks: Payoffs and probabilities Beliefs become especially important in cases where we have to take some action without knowing for sure what the outcome will be. You make many of this kind of choices every day, from the important choices of what to study at university, to more trivial choices like whether to take an umbrella to class. The theory of decision-making in these cases rests on the idea that the evaluation of how good a course of action is depends on • how much the decision-maker values each of the possible but uncertain outcomes of the action and • the decision-maker’s beliefs about how likely each is. Here we introduce a basic concept for decision-making with risk – expected payoffs – that will be used throughout the book. In Chapter 13 we return to the topic of risk including preferences about risk taking per se and the value of insurance. The value of uncertain outcomes: Expected payoffs There are two possible but uncertain outcomes of the action "take an umbrella to class," namely, "keep dry walking home in the rain" and "carry the umbrella to and from class without even opening it, because it does not rain." The feasible actions of the decision maker are just: take the umbrella or not. According to the preferences beliefs and constraints approach, the decision maker assigns numbers indicating how much she values each of the possible four outcomes shown in Table 2.2. These numbers give the ranking of the four possible outcomes: [Don’t take the umbrella, No rain] is better than [Take the umbrella, Rain] and so on. But if they are to provide a framework for making a decision when you do not know for sure if it is going to rain or not, Figure 2.3: Amartya Sen (1933- ). Image Credit: National Institutes of Health, Public Domain. the numbers have to be more than a ranking. They have to indicate how much the actor values each of the possible four outcomes. So for example taking the umbrella when it rains is 5 times better than not taking the umbrella when it rains. We call these numbers the payoffs to each of the four possible outcomes. The likelihood of uncertain outcomes: Beliefs Only one of these two uncertain events will occur. Whether at the end of the day, it turned out to have been a good idea to have brought the umbrella said to be contingent on (meaning: depends on) whether it rains or not. The C ONTINGENCY The payoff to the outcome of a decision is said to be contingent if something affecting the payoff may or may not happen. The payoff in this case is said to depend on a contingency. payoff to the two actions in this case is said to depend on a contingency. The contingency in this case is whether or not it rains and the payoff to taking the umbrella is contingent on (depends on) its occurrence. P ROBABILITY DISTRIBUTION A probability distribution for n contingent outcomes of a decision is a list of non-negative numbers {P1 , P2 , . . . , Pn } that add up to 1. These probabilities express the decision-maker’s degree of belief about the likelihood that each of the of n contingent outcomes will occur. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES Uncertain event (contingency) Action Rain No Rain Take the umbrella 15 8 Don’t take the umbrella 3 20 79 Table 2.2: Two contingencies (rain or don’t rain) and two actions (Take the umbrella, or Don’t). The payoffs correspond to the coincidence of an action and a contingency, so Anoushka receives 15 if she plays Take the umbrella when the contingency is Rain, and she receives 8 if she plays Take the umbrella and the contingency is No rain. When you decide what to study at university before knowing what kind of work you’ll do after, you’re making choices about contingencies too: do you go risky and study in drama, or do you go safe and take in accounting? In this case, the contingencies include the the uncertainty about how good you will be at the field you choose and your chance of getting a job in your field. The theory of decision-making about risky outcomes assumes that a decisionmaker, call her Anoushka, has beliefs about the probabilities (Pi ) that each of the contingencies i = 1, . . . , n will occur. Her beliefs can be based on observation, on empirical studies, guessing, experience, or superstition. They R ISK AND UNCERTAINTYThe term risk is conventionally used in economics to describe situations where the probabilities of the possible outcomes are known. The term uncertainty describes situations where the decision-maker does not know and cannot learn these probabilities. need not be correct. For simplicity we assume contingencies with just two outcomes (like "it rains" or "it does not rain" above). The basic principles of decision-making are the same no matter how many contingent outcomes there are. In this case, we use the symbol P for the probability the contingency occurs, understanding that 1 P is the probability the contingency does not occur. The decision rule: Maximize expected payoffs Often we must take an action prior to the realization of the outcome – you do not know with certainty what will happen, that depend on an uncertain events – called a contingency – that may or may not happen. But you have to make a choice anyway. To take account of the "action now, contingency later" aspect of the decision problem we distinguish between: • Expected payoff : how much the actor values taking the action given her beliefs about the probability that the contingency will occur and • Realized payoff : how much she values the possible outcomes that may happen, that is, after the contingency has been realized ("realized" here means really happened, or actually occurring). The expected payoff of an action is the basis for her choosing one course of action over another: Anoushka chooses the action with the highest expected payoff. Here is how she can calculate expected payoffs. For each contingency, i, and each action she can take, x, Anoushka knows the payoff of taking action x conditional on i happening, which we write as p (x|i). For example, if i is the contingency of rain in the afternoon, and x is the action of taking her umbrella with her in the morning, then her realized payoff is p (x|i) associated with her having the umbrella when it rains. The vertical line H I S TO RY In 1947 John von Neumann and Oskar Morgenstern showed that how much we value some action that we can take, when the outcome of the action is subject to some risky contingency can be expressed as a weighted sum of how much we value the alternative outcomes of our actions that depend on the realization of the contingency, the weights being the probability of each outcome occurring if we take the action. 80 MICROECONOMICS - DRAFT | is read "conditional on", or "given", so p (umbrella|rain) is Anoushka’s payoff to having the umbrella conditional on, or given, rain in the afternoon. For a contingency with two outcomes – numbered 1 and 2 – we have to consider only two payoffs and the corresponding probabilities of each, ((p (x|1), P), (p (x|2), 1 P)). For example, if Anoushka’s payoffs for the four possible outcomes of her actions are as in Table 2.2, and the probability of rain in the afternoon as 0.6, her system of contingent payoffs for taking the umbrella is (15, 0.6), (8, 0.4)). These numbers can be interpreted as follows: since there is a 60% chance of rain, Anoushka has a 60% chance of receiving a payoff of 15 if she takes the umbrella. Further, this means that there is a 40% chance of no rain, therefore, if Anoushka takes the umbrella she has a 40% chance of having a 8 payoff. The expected payoff to an action x given a system of contingent payoffs is the weighted average of the payoffs for each contingency where the weights are the the probability of each contingency being realized if the action x is taken. We abbreviate the expected payoff to choosing x given the probabilities (Px ) of contingencies 1 and 2 being realized as E (px , Px ) = E ((p (x|1), P), (p (x|2), 1 Expected Payoff P)) E (px , Px ) = Pp (x|1) + (1 P)p (x|2) (2.1) Equation 2.1 expresses the fact that the greater the probability of an outcome, the greater its weight in the weighted average calculated by the expected payoff. For example, Anoushka’s expected payoff to taking the umbrella, assuming the payoffs and probabilities above, would be 0.6 · (15) + 0.4 · (8) = 9 + 3.2 = 12.2. Calculating expected payoffs with probabilities is essential to understanding strategic interactions, such as the games we introduced in Chapter 1. But in games – that is strategic interactions with other people – the contingencies are the strategies chosen by the other player, not something like whether it rains. Checkpoint 2.2: Basis of probability assessments a. Imagine that you are rolling two six-sided dice with sides corresponding to one of each of the numbers 1, 2, 3, 4, 5, and 6. You calculate the sum each time you roll the two dice simultaneously, for example, 1 + 2 = 3. Explain why the probability of getting a total of 7 from rolling the two dice is 1/6. b. What is the expected payoff if you get paid $5 for rolling a sum of 6 or 8 on a roll of the two dice and $0 otherwise? c. Go back to Table 2.2, what would Anoushka’s expected payoff to not taking the umbrella be given the probability of rain being P = 0.6? PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES Figure 2.4: Planting in Palanpur: An Assurance Game. Aram’s payoffs are listed in the blue bottomleft corner. Bina’s payoffs are listed in the pink top-right corner. Aram’s best response to Bina’s choice of strategy is indicated by a black dot in the relevant cell, while Bina’s best responses are indicated by hollow circles. Using the dot and circle method you can confirm that the the Nash equilibria of the game (any cells with both a dot and a circle) are (Plant Early, Plant Early) and (Plant Late, Plant Late), with payoffs (4, 4) and (2, 2). The Plant Early Nash equilibrium is Pareto-efficient. The Plant Late equilibrium is not. Bina Early Late Aram Early ● Late 3 4 0 4 0 3 ● 81 2 2 2.3 Expected payoffs and the persistence of poverty In games like the Prisoners’ Dilemma which have a dominant strategy equilibrium, the action that will maximize your payoffs does not depend on what the R E M I N D E R A dominant strategy equilibrium is a strategy profile in which all players play a dominant strategy. other player does, so it does not matter that you do not know what the other will do. But if – like most games – there is not a dominant strategy equilibrium, then your best response depends on what the others do, and we need to take account of this in our decision making rule. We can use expected payoffs to understand the choice of which strategy to play in an Assurance Game, like a farmer’s choice between Planting Early or Planting Late in the Planting in Palanpur game. The game is shown in Figure 2.4 to remind you of the game’s structure. The payoffs in each cell indicate how much the farmer values outcome resulting from the strategy profile given by the particular row and column. As you know, the game has two Nash equilibria: (Early, Early) and (Late, Late). Recall that (Early, Early) is Pareto-superior to (Late, Late). The Plant Early equilibrium is also the payoff-dominant equilibrium. An equilibrium is payoff dominant when no other equilibrium exists that is Paretosuperior to it. The Pareto-efficient Nash equilibrium in an assurance game is payoff-dominant. In our example, Plant Early is payoff-dominant because the payoffs in this equilibrium exceed the payoffs for both players in the Plant Late equilibrium. As we observed in Chapter 1, Palanpur farmers plant late even though a Pareto-superior alternative exists. To see why this occurs, think of what the other player will do as a contingency. We can then say that the degree of belief that other farmers will plant early can be expressed as a probability, P. A farmer believing with probability P that the other farmer will plant early and probability (1 P) that the farmer will plant late is an example of decision- PAYOFF - DOMINANT E QUILIBRIUM An equilibrium is payoff dominant when no other equilibrium exists that is Pareto-superior to it. The Pareto-efficient Nash equilibrium in an assurance game is payoff-dominant. 82 MICROECONOMICS - DRAFT Bina Late Aram Early Early (P) Late (1−P) 4P 0(1−P) 3P 2(1−P) Figure 2.5: Aram’s view of Planting in Palanpur. The figure shows Aram’s payoffs only (and not the payoffs to any other farmers) and his belief about the probability they will occur. Aram’s payoffs are multiplied by the probability of that cell occurring based on the probability that Bina plays that strategy. The other farmers play Plant Early with probability P and Plant Late with probability 1 P. We can calculate Aram’s expected payoffs to each of his strategies by adding the payoffs to a given strategy against each of the other farmers’ strategies. Plant Early: p̂Early = 4 · P + 0 · (1 P) Plant Late: p̂Late = 3 · P + 2 · (1 P). making under risk, since the farmer assigns probabilities to a contingency, in this case, the other farmer’s behavior. We do not explore where these beliefs about probabilities come from, but we can imagine that he farmer will form beliefs based on what other farmers tell him or on the basis of their behavior in past planting seasons. We will include just Aram and Bina in the game, but remember we use only two players to simplify our analysis of what is really a much larger population of many people like Aram and Bina. If Aram believes that the probability of Bina planting early is P we can con- ˆ, M - C H E C K We label expected payoffs as pi which is sometimes written E [p ]. struct his expected payoffs to each of his strategies, each part of which is shown by Figure 2.6. We use a "hat" on a variable to mean ’expected,’ so p̂ reads "p hat." Using these probabilities, Aram’s expected payoff (E (p )orp̂ ) to playing Plant Early is: p̂ = p̂ (plant early)) = Pp (plant early|others plant early) +(1 P)p (plant early|others plant late) Aram’s expected payoff to planting late is: p̂ (plant late) = Pp (plant late|others plant early) +(1 P)p (plant late|others plant late) An expected payoff-maximizing farmer will choose to plant early or late depending on which expected payoff is higher. As Figure 2.6 shows, for Aram, which action this will be depends on the probability that he thinks Bina will plant early. The vertical axis is the expected payoff to each strategy: Plant Early or Plant Late. The horizontal axis is the probability, P, that Bina plants early: from left to right P goes from P = 0 (the Bina plants late with certainty) to P = 1 (the Bina plants early with certainty). The two upward-sloping lines plot the expected payoffs to the two strategies, Plant Early and Plant Late, for Aram the farmer making a decision and how I NDIFFERENCE P ROBABILITY In a two-by-two game, let P be the probability that player A attributes to B playing one strategy and 1 P the probability A attributes to B playing the other strategy. Then Pi is the value of P such that such that player A’s expected payoffs to to playing each of her two two strategies are equal. In this case player A is therefore indifferent between playing the two strategies (which is why we use the letter i subscript.). PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES ^ Expected payoff, π Strategy with higher expected payoff at P = 1 2 is risk−dominant 4 Equal expected payoffs at Pi = 2 3 3 2.67 2.5 r 2 i Expected payoff to ^L Plant Late, π Expected payoff to ^E Plant Early, π P = 1 2 Pi = 2 3 0 1 Probability Bina will play early, P they depend on the probability that he believes Bina will plant early (that is, for each value of P). The blue line graphs the equation for the payoff to the strategy Plant Early which is p̂Early (P) = P · 4 + (1 P) · 0 = 4P. When the probability the other farmer plants early is zero, i.e. P = 0, the payoff to Plant Early is zero. When the probability the other farmer will Plant Early is 1, i.e. P = 1, the payoff to Plant Early is 4. We can draw the expected payoff line for Plant Late in the same way, where the expected payoff to Plant Late is p̂Late (P) = P · 3 + (1 P) · 2 = 2 + P depicted in green, and where pLate (P = 0) = 2 and pLate (P = 1) = 3. We can then interpret the expected payoffs as follows: • Plant Late provides a higher expected payoff for all P < 2/3. • Plant Early provides a higher expected payoff when P > 2/3. • The expected payoffs to the strategies are equal at the indifference probability Pi = 23 (where a farmer is indifferent between Plant Early and Plant Late). The result is that Aram will choose Plant Late as long as he believes that the probability that Bina will Plant Early is less than two-thirds. Bina, facing the identical situation, has the same decision rule: Plant Late unless you think that Aram is going to Plant Early with a probability of at least two-thirds. They will remain poor even though, had they somehow started of both planting 83 Figure 2.6: Aram’s expected payoffs to planting early or late depend on his belief about the probability that Bina will plant early. Aram evaluated the expected payoffs to his strategies based on the probability that Bina will play Early. The indifference probability where the two strategies have the same expected payoff is Pi = 2/3, and the payoff to Planting Late is greater than the payoff to Planting Early for P = 1/2. The intercepts of the vertical axes are the payoffs in the payoff matrix for the planting game in Chapter 1 (Table 2.4). 84 MICROECONOMICS - DRAFT early, they would have been much better off. The poverty trap in which they find themselves is not the result of rudimentary technology or infertile soil. What they lack is the "social technology" that would allow them to coordinate on the Pareto-superior strategy profile, planting early. Their poverty is due to the rules of the game, which make coordination difficult. In this example we have assumed that both Aram and Bina had some idea (maybe a guess) of the likelihood that the other would plant early. They faced risk (they had some information on the probability of the contingent event), but not uncertainty (no information at all). Decision making under uncertainty is especially important in the field of climate change, where there are some contingencies for which there is no way to assign probabilities of their occurrence. M-Note 2.1: Expected Payoffs for Planting in Palanpur To understand the expected payoffs and the indifference probability in the game, we need to answer the following questions: a. What are the expected payoffs to planting early and planting late (using payoff Table 2.4) for a farmer in Palanpur who believes the probability of Bina planting early is P, with 0 < P < 1? b. What value of P leads to an equal expected payoff to planting early and late? We can work out these answers using the following steps: • If Aram plants early and Bina plants early, Aram’s payoff is 4 (with probability P) • If Aram plants early and Bina plants late, Aram’s payoff is 0 (with probability 1 P) • If Aram plants late and Bina plants early, Aram’s payoff is 3 (with probability P) • If Aram plants late and Bina plants late, Aram’s payoff is 2 (with probability 1 P) The expected payoff to planting early is the weighted average of the two contingencies (Bina plays Plant Early or Plant Late) with the weights equal to the probability (P) of Bina playing Plant Early and (1 P) of Bina playing Plant Late: • Expected payoff for planting early p̂Early (P) = 4 · P + (0) · (1 • Expected payoff for planting late: p̂Late (P) = 3 · P + 2 · (1 P) = 4P. P ) = 2 + P. • To find the indifference probability at which expected payoffs are equal: p̂Early (P) p̂Late (P) = • Which is the condition: P = Pi = 23 , the indifference probability. 2.4 Decision-making under uncertainty: Risk-dominance What is the farmer facing uncertainty to do? Economics does not have a very good answer. A two-person risk-dominant equilibrium Economists often use what is called the "principle of insufficient reason" when a player has no information on which to place a probability on some H I S TO RY The "principle of insufficient reason" due to the Swiss mathematician Jakob Bernoulli (1655-1705) states that if we have no information on which to estimate the probability that one of two contingencies will occur, we should consider them to be equally likely. Not everyone finds this satisfactory. John Maynard Keynes found it "paradoxical and even contradictory."4 PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 85 contingency. This principle holds that the farmer who has no information on likely strategy choice of his neighbor will assign equal probability to the two events and hence use the probability P = 12 that the other will plant early. What is termed the risk dominant strategy is that which yields the highest expected payoff when a player attributes equal probability to the two actions of the other player. Using this definition in the Planting in Palanpur game, a farmer who assigns the probability P = 12 to the outcome that the other farmer will Plant Early will himself Plant Late because his expected payoffs are 2 = 12 · 4 to Plant Early, and 2.5 = 12 · 3 + 12 · 2 to Plant Late. Thus Plant Late is the risk-dominant strategy, that is, the strategy that maximizes the farmer’s expected payoffs when P = 1/2. You can confirm this by going back to Figure 2.6: at P = 1/2 the green line (expected payoff to planting late) is above the blue line (expected payoff to planting early). Because this is true for the other farmer as well, both farmers Planting Late is the risk-dominant equilibrium. Planting late in the Planting in Palanpur game is risk dominant because planting early when the other plants late is much worse (you get zero rather than the payoff of two you would have received had you also planted late) than planting late when the other plants early (you get three rather than the four you would have received had you also planted early). Checkpoint 2.3: Risk dominance and the worst case outcome a. Redraw the expected payoff line for planting early with the payoff to planting early when the other plants late to be even worse than shown in the figure, e.g. -2 instead of 0. b. In this case what is the indifference probability? c. What is the least payoff to planting early when the other plants late, that would make planting late no longer risk dominant? A risk dominant equilibrium in a large population Instead of thinking about only two farmers, we can interpret as portraying a population of farmers in a village like Palanpur itself, who all face the same set of incentives for planting early and late. Like Aram and Bina, all the farmers face a coordination problem: doing well if they all plant early and doing poorly if they all plant late. How well each farmer does depends on what the others do, so if a minority of farmers plants early while the majority plants late, then those who planted early while others planted late will have their seeds eaten while the others will get an adequate harvest. The farmers are therefore involved in a many-player coordination problem. We can re-purpose Figure 2.6 such that the horizontal axis is the fraction R ISK DOMINANT STRATEGY The strategy that yields the highest expected payoff when the player attributes equal probability to the two actions of the other player. 86 MICROECONOMICS - DRAFT ^ Expected payoff, π 4 3 2.67 2.5 r 2 i Expected payoff to ^L Plant Late, π Figure 2.7: Fraction of farmers planting early. P is the fraction of farmers playing Plant Early. 1 P is the fraction of farmers playing Plant Late. In the case of the population as a whole, the indifference probability (or in the case of a population of players, the indifference fraction) shown at point i with fraction Pi corresponds to the fraction of the population at which the players are indifferent between the strategies (Plant Early or Plant Late). In the case of the whole population, point i is also the tipping point: when a fraction of the population less than Pi plays Plant Early all farmers will want to play Plant Late; when a fraction of the population greater than Pi plays Plant Early all of the farmers will want to play Plant Early. Expected payoff to ^E Plant Early, π P = 1 2 Pi = 2 3 0 1 Fraction playing Plant Early, P of the population going from 0 to 1 who choose Plant Early, P (reading left to right) as shown in Figure 2.7. Reading the horizontal axis from right to left, it measures the fraction of the population who Plant Late (1 P). The payoff lines in the figure have the same interpretation as before: They are the expected payoffs for any one of the large number of identical farmers in the village. The probabilities translate to population fractions too: • P < 23 : When less than two-thirds of the population choose Plant Early (i.e. more than one-third play Plant Late), the Plant Late strategy will get him a higher expected payoff. At any fraction P < 23 all farmers will Plant Late and all farmers will end up with a payoff of 2. • P > 23 : When more than two-thirds of the population select Plant Early (i.e. less than one third select Plant Late), the Plant Early strategy has a higher expected payoff. At any fraction P > 23 all farmers will Plant Early and all farmers will end up with a payoff of 4. • P = 23 : At two-thirds Planting Early and one-third Planting Late, the expected payoffs are equal. The point at which the expected payoffs are equal is a tipping point as a small change will drive all players to adopt one or the other strategy: Plant Early or Plant Late. Now imagine that as in the village of Palanpur virtually all of the farmers have been planting late year after year (maybe even generation after generation). There would not be much uncertainty about what fraction of the population would plant late the next planting season. Each of the farmers would hold the belief that P is close to zero and as a result they all would plant late, confirming their beliefs. The belief that almost nobody would plant early sustains both the low income of the farmers, and perpetuates the belief itself, which year after year turns out to be correct. T IPPING P OINT An intersection of the expected payoffs to strategies shows a tipping point when a small change in population fractions playing a strategy results in a feedback loop driving the game to one of the extremes, either P = 0 or P = 1. We describe it as a tipping point since a small push either way will "tip" the outcome to one extreme equilibrium or the other. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 87 In the Fishermen’s Dilemma the best outcome for one of the players is the worst for the other, so there is a conflict of interest between the two. And this contributes to the difficulty of finding some way of coordinating so as to avoid over-exploitation of the fishing stock. This is not the problem in the Assurance Game. There is no conflict of interest: All of the Palanpur farmers prefer the outcome when they all Plant Early to any other outcome. Their failure to implement the mutually desired outcome is the result of their inability to coordinate on planting early, for example when all are planting late to make a joint decision to all change to planting early. What may seem to be a minor tweak to the rules of the game under which the farmers are interacting can help them escape their poverty trap. 2.5 Sequential games: When order of play matters When we looked games involving risk and uncertainty, we saw that players could end up selecting risk-dominant strategies that implement Paretoinefficient Nash equilibria. But the game we introduced to model the coordination problem facing Aram and Bina was unlike many real world social interactions, they were total strangers who had no way of coordinating their actions, and they acted simultaneously (or at least, without knowledge of what the other had done.) But it might be that rather than playing simultaneously, they play sequentially. Playing sequentially is a change in the rules of the game; it represents a change in the institutions governing their interaction. We will see that this seemingly small change makes it into an entirely different kind of game possibly even allowing Pareto-efficient outcomes. To see how this could work, suppose the Planting in Palanpur game the Aassurance Game) is now sequential. Aram moves first (he is called the first mover) and Bina moves second. How will Aram reason? He has to think about what Bina will do in response to his planting early or late. He knows that: • Bina’s best response to his planting late is to plant late and the best response to his planing early is to plant early, and • his payoff is greater if they both plant early. So he will announce that he will plant early, and Bina will respond with planting early. Rather than being stuck planting late with a small harvest, they have now solved their coordination failure. How did they manage it? The answer is that the sequential nature of the game gave them a way of acting together even if they had no way of actually coordinating. By looking ahead to what Bina would do later Aram he was able to act so that they would B ACKWARD INDUCTION Backward induction is a procedure by which a player in a sequential game chooses a strategy at one step of the game by anticipating the strategies that will be chosen by other players in subsequent steps in response to her choice. (Induction here means causation.) 88 MICROECONOMICS - DRAFT Aram ● Aram ● Plant early Plant late Bina ● Plant early Plant early ● Plant late Plant early Bina Plant late Bina ● Plant late Plant early ● ● ● ● ● (4, 4) (0, 3) (3, 0) (2, 2) (4, 4) (a) Full Game Tree Aram ● Plant early ● Plant late Bina Plant early Plant late Bina Plant late Plant early Bina Plant late Plant early Plant late ● (0, 3) (3, 0) (2, 2) (b) Aram’s Reduced Choice of Actions together implement the single Pareto efficient outcome. What Aram did is called backward induction, which is is a procedure by which a player in a sequential game chooses a strategy at one step of the game by anticipating the strategies that will be chosen by other players in subsequent steps in response to her choice. To see how this works when order of play matters (and where backward induction gets its name), instead of using a payoff matrix (or as we have done for normal form simultaneous move games) we will use a game tree and refer to the game as being an extensive form game. Game Trees have the same basic structure as a normal form games represented by a payoff matrix – they show the strategy set and the payoffs associated with each strategy profile – (4, 4) (1, 3) (3, 1) (2, 2) (c) Fully Solved Game Figure 2.8: Game tree of the Planting in Palanpur (assurance) game. 2.8 a presents the full game tree for both players. 2.8 b shows the reduced set of action that Aram considers while using backward induction to solve the game. 2.8 c shows the solved game tree with the arrows indicating the path to the Nash equilibrium (Plant Early, Plant Early). Aram’s actions are shown by the blue branches and Bina’s by the red branches. Aram’s actions are reduced because he has projected forward in time and used backward induction to work out what Bina will do: planting early if he plants early and planting late if he plants late, therefore reducing Aram’s choices to a payoff of 4 if he plants early and a payoff of 2 if he plants late. So backward induction leads to the Nash equilibrium of the game being (Plant Early, Plant Early) with payoffs (4,4) except that the tree-like structure tells us something about who moves when; and a strategy profile is now a path through the branches of the tree. In the game tree structure, the players move in sequence, with the player on the top of the tree moving first and the player at the bottom moving last. A game tree for the sequential version of the Planting in Palanpur game is shown in Figure 2.8. Aram is the first player, so he is at the top of the game tree. Bina is the second player, so she is shown acting after Aram. Each player’s action – planting early or planting late – is shown alongside a branch of the tree to indicate which action the player chooses as they move along that branch. Each player’s payoffs are shown at the end of a branch of the game tree that indicates a specific path to that end point, Aram planting early, then Bina planting early; Aram planting early, then Bina planting late; and so on. The payoffs correspond to those we used in Chapter 1. Because of the branching tree-like structure of the figure there is only one path to each of the end points. In Figure 2.8 on the left-hand side we have the full game tree, showing all the potential payoffs for the game. Bina is the second-mover and she needs to decide what to do at each point where she could move. If Aram plants early, Bina can get a payoff of 4 for planting early, or a payoff of 3 for planting late. So if Bina is self-interested, then she will plant early when Aram plants early E XTENSIVE F ORM G AME A game portrayed by a game tree in which the sequence of actions by the players is made explicit. The player at the top of the tree moves first, with subsequent players moving in sequence after the first player. Payoffs are shown at the end of the game tree in player order, e.g. (Player A’s Payoff, Player B’s Payoff). Refer to Chapter 1 for the definition of normal form games. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES (4 > 3). Bina also has to make a choice between her actions if Aram plants late. Bina can get a payoff of 0 if she plants early given Aram planting late or a payoff of 2 if she plants late given that Aram plants late (2 > 0). So if Bina is selfinterested, she will plant late when Aram plants late. We now know what Bina will do, but what will Aram choose to do knowing this? Using backward induction and having a belief that Bina is self-interested, Aram will have a choice between a reduced set of payoffs, shown in the central panel of Figure 2.8: either 4 if he plants early or 2 if he plants late. So if he is self-interested, then he will choose to plant early. As a result, the only Nash equilibrium of the game is (Plant Early, Plant Early) with payoffs (4, 4). Checkpoint 2.4: Back in Palanpur 89 E X A M P L E The timing of a sequential game does not depend on the exact actions being taken in that sequence, but can depend on commitments to those actions being taken. For example, in the second quarter of the year a company could commit to the pricing strategy it will follow in the third quarter and if they believed the company’s commitment, then other companies have to respond to that commitment, even if it’s not the third quarter yet. Similarly, a professor commits to a policy in her syllabus even if her students haven’t written a midterm exam or solved a problem set yet. The student must respond to the professor’s committed actions and work out what he will do in response to her commitment. To design the syllabus the professor using backward induction thought through what a student would do in response to her commitments in the syllabus. Making the game sequential solved the problem for Aram and Bina. But would that work for the couple of hundred families in Palanpur? Suppose some order of play was determined and that the first family had announced that they would plant early. Would the second family then follow? And the third? Explain why or why not? 2.6 First-mover advantage in a sequential game Being first mover did not give Aram any particular advantage in the Planting in Palanpur Game, it just allowed him and Bina to coordinate on the Paretoefficient Nash equilibrium. The result would have been the same had Bina been first mover. But sometimes it is advantageous for a player to move first; this person then has what is called first-mover advantage. Think about the Disagreement Game from Chapter 1. Recall that two players, Aisha and Ben, have a disagreement over which (or perhaps both) of them should study to improve the language spoken by the other. Both prefer when they are good at speaking some language. But, Aisha prefers that it be Swahili and Ben prefers that it be English. What happens in this game when Aisha is the first mover rather than when they both move simultaneously? Considering the game tree in Figure 2.9, we can solve the game by backward induction and see that the Nash equilibrium of the game is (Stick to Swahili (for Aisha), Improve Swahili (for Ben)) with payoffs (4, 2). The outcome (Improve English, Stick to English) which was one of two Nash equilibria in the simultaneous version of the game is no longer a solution in the sequential version of the game. Aisha does better as a first-mover because she obtains her F IRST- MOVER ADVANTAGEA player who can commit to a strategy in a game before other players have acted is a first mover. This limits the outcome of the game to a strategy profile made up of his chosen strategy and to the other players’ best responses to it, which may result in higher payoffs for the first mover. This is called first-mover advantage. 90 MICROECONOMICS - DRAFT Aisha Aisha Improve English Stick to Swahili Ben Improve Swahili (4, 2) Ben Stick to English (0, 0) Improve Swahili (0, 0) Improve English Stick to Swahili Ben Stick to English (2, 4) Improve Swahili (4, 2) (a) Full Game Tree preferred outcome. Ben would have benefited in the same way had he been first mover. The reason why being first mover gave Aisha an advantage is that the normal form game has two Nash equilibria – one preferred by Aisha and the other by Ben. In the sequential game the first mover determines which of the two Nash equilibria will occur. Once Aisha has moved and has established that she will Stick to Swahili (and not try to Improve English), Ben needs to choose his action. Ben needs to take Aisha’s move as given. He must therefore choose his best response to Aisha choosing Stick to Swahili. Given that he would like to communicate with Aisha, his best response is to Improve Swahili. We have not asked why Aisha rather than Ben was first mover in the Disagreement Game and why it was Aram rather than Bina in Planting in Palanpur. We showed only that differences in a person’s position in the game gave them advantages. In this case Aram and Aisha – the first movers – had strategy sets that gave them the capacity to commit to a strategy in advance. Aram’s first mover status did not allow him to benefit at Bina’s expense; but this was not the case with Aisha, her first mover status gave her an advantage over Ben. First movers in a modern economy are more like Aisha: • Employers: they commit to the wage, job requirements and working conditions; workers – actual and prospective – best respond to that. • Banks and other lenders: they set to the interest rate, repayment schedule and other aspects of a loan contract. Borrowers and would be borrowers best respond to that. • Owners of major companies: in the U.S. Walmart, Amazon, Apple – com- Ben Stick to English Improve Swahili (0, 0) (0, 0) Stick to English (2, 4) (b) Solved Game Tree Figure 2.9: Game tree of the Language (Disagreement) game. The left-hand side presents the full game tree for both players. The right-hand side figure shows the solved game tree with the arrows indicating the path to the Nash equilibrium (Stick to Swahili, Improve Swahili). Aisha’s actions are shown by the blue branches and Ben’s by the red branches. Aisha’s actions are reduced because she has projected forward in time and used backward induction to work out what Ben will do: Stick to Swahili if he plays Improve Swahili and Improve English if he plays Stick to English, therefore reducing Aisha’s choices to a payoff of 4 if she plays Stick to Swahili and a payoff of 2 if she plays Improve English. So backward induction leads to the Nash equilibrium of the game being (Stick to Swahili, Improve Swahili) with payoffs (4,2). The outcome favors Aisha over Ben and therefore conveys a first-mover advantage. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 91 mit to prices and delivery schedules. Consumers best respond. The fact that people occupy different positions in our economy – Employers and workers, lender and borrowers – interacting under rules of the game that give some first mover status and other special advantages is an important part of the explanation of inequality of wealth and income as we will see in Chapters11,12, 13, and 15. Checkpoint 2.5: Ben has the first-mover advantage a. Consider the sequential Disagreement Game shown in Figure 2.9. Re-draw the game tree, but with Ben as the first mover rather than Aisha. Show that (Improve English, Stick to English is the Nash equilibrium of the game. b. Assuming that the payoffs in the Disagreement Game are in hundreds of dollars and that you are Ben, how much would you pay for the privilege of being first mover a) if otherwise Aisha would be first mover, and b) if the game were to be played simultaneously (so that there is no first mover)? 2.7 Social preferences: Blame Economic man? The characters in our economics episodes – Aisha and Ben, Aram and Bina – care exclusively about their own payoffs. For them a "best response" is simply a "best-for-me response." Is that why they have had difficulty overcoming the coordination failures they face? The answer, we will see, is that being concerned about how your actions affect others will help to address coordination failures, but will not be sufficient. Homo economicus or "economic man" is the term economists have used to designate an entirely self-regarding and amoral actor, a person who is not motivated by either a concern for others, or a desire to conform to any ethical principles. The term is often put in italics to parallel the biological terminology for a species (like Homo sapiens). Homo economicus, however is a fictional character or ideal type representing one possible variety of human behavior. Models based on Homo economicus have provided predictions about behavior that are borne out by empirical studies that range from how American windshield installers and Tunisian sharecroppers respond to different work incentives to the effect of taxes on cigarette consumption. But, as we shall see, Homo economicus is not an accurate depiction of how people behave: • People volunteer for fire fighting, delivering food to the sick during a pandemic, and other dangerous but socially beneficial tasks, and contribute substantial sums to charity. • People participate in joint activities such as strikes or protests even knowing that their individual participation is unlikely to affect the success of H I S TO RY The idea of basing economics on the assumption that people are entirely self-regarding –"solely as a being who desires to possess wealth" goes back to the last of the great classical economists, John Stuart Mill author of Principles of Political Economy (1848), considered to be the first economics textbook in the English language. He considered this view of people to be "an arbitrary definition of man."5 92 MICROECONOMICS - DRAFT the event and that, if successful, the benefits would be widely shared, not confined just to those people participating in the protest. • People donate blood to strangers. • In public opinion polls and in voting, people support taxes that transfer incomes to the poor even when they are sufficiently rich and unlikely ever to benefit directly from these policies. Motivated by these and similar observations a augmented by controlled experiments about human behavior (that we will review below), economists have revised our assumptions to recognize that people are capable of ethical, generous, and other motivations as well as self-regarding motives. This is important because as you learned in the first chapter, coordination failures occur because we fail to take adequate account of the effect that our actions have on others. Our concern for others can help to internalize these external effects whether it be our willingness to curb our carbon footprint or willingness to protest for causes whose benefits would be widely shared. But coordination failures cannot be blamed entirely on people seeking to maximize their own payoffs. Think again about the real farmers in Palanpur, all planting late when they could all do better if they all switched to planting early. Suppose one of those farmers was deeply concerned about the poverty of his entire village, and wished to improve living standards for everyone. He could not do this by individually planting early. Now suppose that every villager shared his concerns for all members of their community. Each one would know that their own decision to plant early would change nothing (except that their seeds would be eaten by the birds). What has captured the people of Palanpur in a poverty trap is not that they care only about their own harvest (they surely care about others’), but their inability to come to a common agreement to plant early. Their poverty stems from a problem of institutions, not motivation. To understand individual behavior and its social consequences we need an approach that allows for the full range of human motivation. Checkpoint 2.6: Homo economicus goes to the polls a. Given that it costs time to cast a vote (going to the voting station, standing in line, and the opportunity cost of your time), do you think a person with Homo economicus preferences would vote in most elections? Why or why not? b. In what circumstances do you think someone would someone with the preferences of a Homo economicus vote? While answering these questions, think about the beliefs the person with Homo economicus preferences would have about the probability his vote will be important to the outcome of the election. R E M I N D E R Remember that in Chapter 1 we saw how ’internalizing the external effects’ means getting people to pay for the external costs they imposed on others and this resulted in the fishermen choosing to cooperate and fish less in the Fishermen’s Dilemma. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 93 And so, while Homo economicus is among the kinds of actors this approach considers, there are other characters, representing other sides of human behavior such as generosity, fairness, reciprocity and spite. What these four aspects of behavior have in common is that they are otherregarding: the outcomes that the actor considers in choosing an action include things experienced by others, not just outcomes affecting the actor herself. Here are some common forms of other-regarding preferences: • Those with altruistic preferences, such as basic generosity, are motivated to help others even at a cost to themselves, they place a positive value on the well-being or payoffs of others. • Inequality-averse or fairness-based preferences motivate people to seek to reduce unjust or unfair economic differences even if the actor is herself a beneficiary of these differences. • A person with reciprocal preferences is motivated to help others who have themselves behaved generously or upheld other social norms, and also to punish those who have treated others badly. • Spite and ‘us versus them’ distinctions that place a negative value on outcomes experienced by others, often motivate hostility towards members of religious, racial, ethnic and other groups. Therefore a negative outcome another person experiences, can result in a positive value for someone who feels spiteful. The term "social preferences" is used to describe all types of other-regarding preferences. Checkpoint 2.7: Social Preferences & Social Norms a. Give an example of a preference you have that is not self-regarding. b. Can you think of any social norms that lead you to act in an other-regarding way? c. Suppose that Aram and Bina (in the Planting in Palanpur Game) were of different religions between which there is hostility, so that each would gain some pleasure from the misfortunes of the other. Can you show how this could change the game so that instead of having the Pareto-efficient mutual early planting as one of its two Nash equilibria, it becomes a prisoners dilemma with planting late as the dominant strategy equilibrium. 2.8 Experiments on economic behavior Suppose you wanted to know if someone has altruistic preferences. How would you find out? Would you ask her? Well, that could provide some information, but merely asking might not be entirely convincing , because many INEQUALITY AVERSION is a preference for more equal outcomes and a dislike for both disadvantageous inequality that occurs when others have more than the actor and and (to a lesser extent typically) advantageous inequality that occurs when the actor has more than others 94 MICROECONOMICS - DRAFT people would like others to think they are altruistic even when they are not, so they might lie. What about observing her behavior and comparing her behavior to how others behave? Such observation might be informative, but if we see people behaving differently that could be because the people we observe have different beliefs or different constraints, not different preferences. Economists use experiments to study preferences because at least ideally this allows us to control for (hold constant) the constraints and beliefs of the individuals to focus on the nature of preferences. Experiments allow economists to implement the ceteris paribus – all else equal – assumption that we think is so important when we are trying to identify causes and consequences of some change or difference. To understand how common different types of preferences are, and how they affect our behavior, economists use laboratory experiments in which subjects C ETERIS PARIBUS is a Latin term that means "other things equal." When we held another player’s strategy constant in Chapter 1 to find a player’s best response we were using the ceteris paribus assumption. Similarly, when we use calculus and mathematically hold other variables constant we are employing the ceteris paribus assumption. play games designed to elicit the nature of their motivations. Experiments play a central role in science: they allow predictions made from theories to be tested empirically. This has has been done, for example, with the prediction that players in a Prisoners’ Dilemma experiment choose the dominant strategy equilibrium. (You will see what happens in this experiment below.) In some sciences, experimenters can control almost all relevant conditions in their environment, the laboratory. Their subjects can be anything from yeast cells for a biochemist, to fruit flies for a zoologist. In economics, however, our subjects are people asked to make decisions or choices in the experiment. It is much harder to control for the various fac- F AC T C H E C K Behavioral experiments are a recent addition to economists’ empirical tool kits; but they have been used in psychology for almost a century and a half. The main innovations that economists have made to experimental social science are the use of game theory to clarify the role of beliefs and preferences and nature of incentives and the common use of monetary payoffs. tors that affect human behavior than to control the chemical environment of a colony of yeast cells. Experimental evidence carries little weight unless the experiment can be replicated, that is, repeated by different researchers reaching the same results. We use a specific vocabulary when we talk about behavioral experiments in economics. The following terms will come up often: • Subject/participant: A subject or participant is a person who participates in an experiment. • Endowment: The endowment is an initial amount of money or tokens later converted to money that subjects receive at the beginning of the experiment, and later make decisions about in the experiment. • Incentives: The fact that players stand to win material rewards in varying degrees depending on how they play the experimental game means that the experiment mimics many real economic interactions.. REPLICATION When other researchers independently repeat an experiment and reach the same results they have replicated the experiment. When an experiment can be replicated we know that its results are reproducible. Science is founded on reproducible evidence. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES • One-shot vs. repeated: A one-shot experiment occurs once and subjects make one decision in the experiment as a whole and are paid for that one decision. A repeated experiment involves subjects making repeated decisions often with information about the play of others on previous rounds, sometimes with the same other subjects in a group or sometimes with different subjects. Here is an example of the importance of using results from experiments to test a theory. Subjects with self-regarding preferences are predicted to defect in a one-shot Prisoners’ Dilemma game because defection is the dominant strategy. But in Prisoners’ Dilemma experiments, the proportion of players who cooperate rather than defect is commonly between 40 and 60 percent.6 This means the prediction based on the assumption that people are entirely self regarding was borne out for some but far from all of the subjects. The finding therefore provoked some rethinking of the predictions based on the assumption that people are entirely self-regarding. Many subjects prefer the mutual cooperation outcome and are willing to take a chance on the other player also not defecting, rather than the higher material payoff they can obtain by defecting when the other cooperates. When subjects defect, experimental evidence suggests it is because they dislike being taken advantage of, not because defection is the payoff maximizing strategy independently of the other participant’s actions. 2.9 The Ultimatum Game: Reciprocity and retribution Observing substantial levels of cooperation in the Prisoners’ Dilemma game was a shock to the standard Homo economicus assumptions. But the experiment that has sparked the perhaps the greatest reconsideration of the Homo economicus model is the Ultimatum Game. Here is the game with its basic treatment: • Subjects are anonymously paired for a "one-shot" interaction with another person. • The role of “Proposer”, is randomly assigned to one of the subjects; the other is then the “Responder”. • The Proposer is given an endowment, the "pie" (e.g. $10), by the experimenters and the Responder knows the size of the pie. • The Proposer then proposes how to divide the endowment between Proposer and Responder, transferring to the Responder any amount between nothing and the entire endowment, e.g. the Proposer chooses to keep $ 8 and give $2 to the Responder. 95 96 MICROECONOMICS - DRAFT Player A Player A ● Offer (8,2) split Offer (5,5) split Player B ● Accept Offer (8,2) split ● Reject Player A ● Accept Player B Reject ● ● ● ● (8, 2) (0, 0) (5, 5) (0, 0) ● Offer (5,5) split Player B ● Reject Accept (8, 2) (a) Full Game Tree Offer (8, 2) split (0, 0) Player B Accept Accept (0, 0) (8, 2) (b) Self-regarding players • If the Responder accepts the proposed division, the Responder gets the proposed portion, and the Proposer keeps the rest. • If the Responder rejects the offer both get nothing and the game ends. Figure 2.10 presents a game tree for a variant of the Ultimatum Game where the Proposer chooses between two offers: divide the pie equally and each person gets $5 for an outcome (5, 5) or keep $8 and offer the Respondent $2 for an outcome of (8, 2). The Responder then chooses whether to accept or reject the offer. The payoffs to each player are listed in the order of play (Player A, Player B), so (8, 2) means Player A gets 8 and Player B gets 2. If the Proposer cares only about her monetary payoffs in the game and believes that the Respondent is similarly self-regarding, then the Proposer (Player A) will reason backwards as follows: • Player A predicts that the Responder (Player B) will accept the offer of $2 (because A believes that B is also self-regarding and because $2 is greater than his fallback of $0 which is what he gets if they reject the offer. • A will propose the (8, 2) split to maximize A’s payoff. • The Responder (B) will accept. 2.10 b 2.10 shows how the game unfolds if Player A and Player B are both entirely self-regarding and maximize their monetary payoffs. Player B will then always prefers a positive money amount over zero, and so they will never reject a positive offer. Player A knows how Player B will respond, and therefore, has a choice between a payoff of 8 and a payoff of 5; they prefer 8 and so offer a split of (8, 2). So if the two player’s are self-regarding then backward induction leads to the Nash equilibrium of the game being (Offer (8, 2) Split, Accept) with payoffs (8, 2). But A, the Proposer, who has followed through this reasoning now understands that she could have offered B the smallest possible positive amount, say $1, and that the offer would have been accepted. Figure 2.10 c 2.10 Player B ● Reject (5, 5) Offer (5, 5) split Player B Reject (0, 0) Accept (5, 5) Reject (0, 0) (c) Reciprocal Player B Figure 2.10: Game tree of the ultimatum (bargaining) game. 2.10 a presents the full game tree for both players regardless of type. Player A is the Proposer and their actions are shown by the blue branches. Player B is the Responder and their actions are shown by the red branches. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 97 shows how the game might be played if Player B is a Reciprocator, that is, Player B cares both about monetary payoffs and also about reciprocating how Player A treats them. In this case, Player B views an offer of (8, 2) as unfair or demonstrating bad intent, and they would rather get a payoff of zero dollars than be treated poorly, so they would reject. If, on the other hand, Player A offers (5, 5) then Player B views that as fair or demonstrating good intent and they would prefer a payoff of 5 in that context to a payoff of 0, so they would accept. Player A prefers a payoff of 5 to a payoff of 0 (this is true regardless of whether player A is self-interested or reciprocal) and so the Nash equilibrium of the game is (Offer a (5, 5) Split, Accept) with payoffs (5, 5). The Ultimatum Game has been played anonymously for real money in hundreds of experiments with university student subjects and other populations – businessmen, fishermen, farmers, civil servants – in all parts of the world.7 The prediction based on the assumption that people are entirely self-regarding and believe that others are too invariably fails as a description of how people F AC T C H E C K Did the subjects not understand the game? It is not that complicated a game, and later experiments in which subjects played the game many times with different partners showed this wasn’t true. Their behavior remained consistent with the one-shot experiments and their results continued to be reproduced with many people making 50-50 splits (or nearly so) and rejecting low offers. behave. For example: • Modal offers – the most common offers in the experiments – are typically half of the pie, and average offers generally exceed 40 percent of the pie, and • Offers of 20 percent of the pie or less are often rejected with frequencies; people in the position of Responder choose to reject and get zero rather than accept and get a payoff of, say, $2 offered from the Proposers $10 pie. These rejections of small but positive offers from the Proposer are interpreted as evidence for reciprocity motives on the part of the Responder. Why? Because the Responder is willing to pay a price (giving up a positive payoff) to punish the Proposer for making an unfair offer (an offer the Responder considers too low). Responders apparently consider a low offer to be a violation of a norm of fairness, and a person with reciprocal preferences responds by depriving the proposer of any payoffs at all. Explaining the behavior of Proposers is more complicated. The outcomes of the experiments are not sufficient to say whether the large number of even splits (and other seemingly fair or near-fair offers) is explained by adherence to fairness norms or altruism by the Proposer or to self-regarding preferences informed by fear that the Responder will reject an unfair offer. The evidence for reciprocity motives therefore, comes from the Responders’ behaviors, not the Proposers’ behaviors.9 F AC T C H E C K Some have suggested that the results were due to the relatively low stakes in the game, such as the $10 mentioned earlier. But subsequent experiments conducted among university students in Indonesia for a ‘pie’ equal to three months average expenditures replicated the results as did experiments with U.S. students with a ’pie’ ranging in size up to $100. Evidence from France showed similar behavior by proposers with stakes ranging from 40 French francs ($7.20) to 2000 French francs ($360) (this was prior to the adoption of the Euro). A further study in India observed stakes that varied by a magnitude of over 1000: From 20 rupees ($0.41) to 20,000 rupees ($410) as the stakes. 8 sparked curiosity among a group of behavioral scientists: Was this simply an odd result, perhaps due to the unusual circumstances of the experiment, or had Henrich tapped real differences, perhaps reflecting the distinct 98 behavioral MICROECONOMICS - DRAFT Figure 1. provides some comparative information about the societies discussed here. In selecting these, we included societies both sufficiently similar to the Machiguenga to offer the possibility of replicating the original Machiguenga results, Locations of the 15 small-scale societies. BEHAVIORAL AND BRAIN SCIENCES (2005) 28:6 2.10 A global view: Common patterns and cultural differences Anthropologists and others were surprised that the results of experiments with the Ultimatum Game have been so similar across the many countries in which they have been conducted. One observed that in virtually all of the early experiments the subjects were from WEIRD countries, meaning Western, Educated, Industrialized, Rich, and Democratic.10 A team of anthropologists and economists (including one of your current authors) designed a series of experiments to explore whether the results reported so far are replicable in societies with quite different cultures and social institutions and whether results differed across the different societies.11 These societies included hunter-gathers, herders, and farmers (some using modern methods, others not even having cattle, horses, or plows). In their Ultimatum Game experiments the pie was substantial, approximately a day’s average wages or other income. Figure 2.11 shows the location of the 15 small-scale societies around the globe. The team was wondering if they would find cultural differences, and they found them. Among the Au and Gnau people in Papua New Guinea offers of more than half of the pie were common, and many of these high offers were rejected. In fact Responders among the Au and Gnau peoples were as likely to reject a offer of much more than half as an offer of much less than half. 799 Figure 2.11: Small-scale societies where the Ultimatum Game experiments were conducted. A map of the world showing the locations of the small-scale societies where the Ultimatum Game experiments were conducted. Source: Henrich et al. (2005). PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES Though this seemed odd to the economists on the team, it did not surprise the anthropologists who study New Guinea. They know that people in New Guinea compete with each other to see who can give more or better gifts. Gift-giving conveys status in their society and people use giving gifts as a way to obtain status over others. Refusing a gift suggests that you are not subordinate to the gift-giver, while accepting it means their status is higher than yours. By contrast, among the highly individualistic Machiguenga slash and burn farmers in Amazonian Peru, almost three quarters of the offers were a quarter of the pie or less and there was just a single rejection, a pattern strikingly different from other experiments. The Machiguenga came as close to acting like Homo economicus as any population yet studied. Even among the Machiguenga, however, the mean offer was still 27 percent of the pie, more than the zero we’d expect if they all were consistently self-interested. The researchers who analyzed the experiments in the 15 small scale societies made the following conclusions: • Although behaviors vary greatly across societies, not a single society approximated the behaviors that would be observed if everyone cared only about their own payoffs and believed others were the same. • Between-society differences in behavior seem to reflect differences in the kinds of social interaction people experience in everyday life. Here is some evidence that the experimental game behavior reflected the lived experiences of the people. • The Ache hunter gatherers in Paraguay share meat and honey equally among all group members. Ache Proposers contributed half of the ‘pie’ or more. • Among the Lamalera whale hunters of Indonesia, who hunt in large crews and divide their prey according to strict sharing rules, the average proposal was 58 percent of the pie. Given the evidence from small-scale societies like the Lamalera and the Ache, we might ask whether we find other-regarding behavior in real-world situations elsewhere in the industrialized world. A different team of researchers were interested in exactly this question and designed an experiment that mirrors a real life dilemma: what would you do if you found a wallet someone had lost: would you return it? The team distributed a total of 17,303 "lost" wallets some with money in them some without, in 355 cities across 40 countries.12 In each country, the researchers targeted big cities to ensure that there was a good sample of subjects (and to ensure anonymity). Using transparent wallets with a business 99 MICROECONOMICS Country 100 - DRAFT Figure 2.12: Wallets with details of their owners were more likely to be given back to their owners when they contained money than when they did not. The "Reporting Rate" is the fraction of wallets that were "returned" Switzerland Norway Netherlands Denmark Sweden Poland Czech Republic New Zealand Germany France Australia Spain Russia Canada Argentina Israel Portugal USA UK Italy Chile Brazil South Africa Thailand Mexico India Turkey Ghana Indonesia Malaysia Kenya Kazakhstan Morocco Treatment Money No Money 20 40 60 80 Reporting Rate card, grocery list, key and cash, the researchers could check how many people contacted the "owner" of the wallet given in the email address listed on the business card to return the wallet. Before reading on, ask what you think would happen in your community: how many people would try to return the wallet? Would more people return the wallet if it had money in it, than if it did not? The results of people’s choices are shown in Figure 2.12. Though there were differences across countries, with just one exception among the 33 countries people were more likely to contact the "owner" if the wallet contained money ($13.45, the treatment) in it than if it did not ($0, the control). In a subset of cases – in the US, UK, and Poland – the researchers added a treatment with even more money in the wallet ($94.15). With a really substantial sum of money in the wallet, people were as likely, if not more so, to contact the listed email address on the business card in the wallet. In interpreting the results keep in mind that the countries differ greatly in how much an additional $13.45 would make to a person standard of living. Per capita income in the richest countries in the sample (Norway for example) are ten and even in some cases 20 times the per capita income in others (Kenya for example), even when account is taken of the differing purchasing power of each national currency at domestic prices. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES The evidence from both the Ultimatum Game and the wallet experiments suggests two important take-aways: • Culture matters: people from different parts of the world live by different social norms and mutual expectations – what we can loosely call "culture." People from different cultures differ in what they consider fair offers and whether think it’s acceptable to make a self-regarding offer. They also differ substantially in whether they will return a lost wallet. • People are similar in many important respects: people across the world have other regarding motives including altruism, fairness, and reciprocity. In the "lost wallet" experiment in most countries a substantial fraction of people attempted to return the wallet. The Ultimatum game and the lost wallet experiment provide valuable information, but they are lacking in one respect. Most of the coordination problems we face involve large numbers of people, like the people of the world making decisions about their carbon footprint, or owners of businesses across the entire economy deciding whether or not to invest, or the herders placing more cows on the commons, or the people deciding whether to drive to work, or the farmers of Palanpur deciding when to plant. 2.11 The Public Goods Game: Cooperation and punishment A public good is one which more people can enjoy without reducing the amount available to others, and from which others cannot be excluded from access to the good. An example is global climate: it is experienced by everyone, and efforts to address the problem of climate change contribute to a public good: that is a more sustainable environment. Another example is the rules of calculus: if you learn how to differentiate that does not deprive others of the knowledge of the same rules of differentiation. This sounds like a good thing. But there is a problem. Why do people produce or contribute to the provision of a public good? If nobody can be excluded from enjoying the good, its hard to see how it would be possible to make money by providing it. (Imagine trying to make a living by selling the rules of calculus!) We return to the problem of public goods in Chapter 5 It shares with the Prisoner’s Dilemma Game the feature that everyone could do better if they agreed on a common course of action (i.e they all contribute) but the dominant strategy for a self-regarding player is not to contribute. For this reason a Public Goods Game is sometimes called an n-person Prisoners’ Dilemma because it has the same incentive structure. 101 102 MICROECONOMICS - DRAFT Rules of the Public Goods Game The Public Goods Game experiment is designed to understand how people will play in a game with this structure. Here are the rules of the game: • n players are each given an endowment of z. • Each player simultaneously selects an amount ei , 0 ei z to contribute to the public good (think of ei as the player’s "effort" in contributing to the public good). • The amount of the public good produced depends on the level of contributions. For example it could be half of the sum of all of the contributions. In this case the average productivity of contributions would be one-half. • Each player, regardless of whether they contribute or not, obtain the entire benefit of the total amount of the public good produced. , As a result of the rules, each player’s payoff can be read as follows: Own payoff = Endowment Contribution + Average productivity ⇥ Total Contributions Figure 2.13 illustrates the benefits of the public good minus the costs of contributing to a public good in a 4-person public goods game. In the version of the game we depict, they can each contribute $10 or $0: which we call "Contribute" or "Don’t." Now compare how a player does if they Contribute (red line) or Don’t (blue line) if they are the only one who contributes, or there are 1, 2, or all 3 others contributing. You can see that in every case she will earn higher payoffs by not contributing. Therefore, if all players are self-regarding, the dominant strategy equilibrium is Pareto-inefficient and an alternative outcome, full contribution by all, which is not a Nash equilibrium, is Paretoefficient. A self-regarding player who cares about only their own costs of contributing and the benefits they get from the public good will choose to contribute nothing (analogous to Defection in the Prisoners’ Dilemma). And this is the case no matter how much or little the other players contribute. So contributing nothing is the dominant strategy for each player. And, like in the Prisoners’ Dilemma, everyone contributing nothing is the dominant strategy equilibrium. As a result economists expected that when this game is played for real money that no player would contribute. They were in for a surprise. M-Note 2.2: Why the dominant strategy in the Public Goods Game is to contribute nothing Payoff net of endowment, $ PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES Figure 2.13: A 4-player public goods game with choices to contribute or not. Each player can play either Contribute or Don’t contribute, and as there are 4 players, this means that the number of others contributing can be any of the numbers 0, 1, 2, or 3 players playing either of the strategies. Playing Don’t contribute yields a higher payoff for the player regardless of how many players play Contribute or Don’t. Therefore, Don’t contribute is a strictly dominant strategy. 15 Contribute Don't contribute 10 5 0 −5 0 1 2 3 Number of others playing Contribute In Figure 2.13, there were four players and we limited their actions to either contributing $10 or contributing $0, but in a standard public goods game players can contribute any amount up to and including their entire endowment (such that e1 = z). In the public goods game in which players can contribute any amount from their endowment, a player’s payoff is given by Equation 2.2: pi = z 103 ei + M  e j for j = 1, . . . n (2.2) j As earlier, we can break down this equation: • z is the endowment of money the player receives from the experimenters. • ei is the contribution a player makes at a cost to themselves. • M is the multiplier, or the average productivity of contributions < 1. •  j e j is the total amount contributed by all players. Whatever the other players do, you can see from Equation 2.2 if you differentiate p with respect to hatei that for person i contributing one unit (say, penny) more changes the her payoff by 1 + M which is the cost of contributing minus the public good that the contributor herself enjoys as the result of her contribution. So as long as M < 1 contributing anything reduces the contributor’s payoffs. This is why not contributing is the dominant strategy Checkpoint 2.8: Two-action Public Goods Game a. Draw a payoff table with two players, A and B, playing the Public Goods Game. Limit their actions to e = 10 and e = 0 with M = 0.5. Check which is the dominant strategy and explain why. What happens if M = 0.75? b. Revise your payoff table and check what would happen if the strategies were e = 1 and e = 0 with M = 0.5? Would anything change? What happens if M = 0.75? c. Think about the condition M < 1 < Mn. Why must this be true for the game 104 MICROECONOMICS - DRAFT Treatment with punishment without punishment ● ● ● Average Contributions 12 ● ● ● ● ● 10 ● ● ● 8 ● ● ● 6 ● ● ● ● ● 4 ● ● 2 ● 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Periods to be an n-person Prisoners’ Dilemma game? (Hint: Think about what would happen if it were not true. What would happen if M > 1? What would happen if Mn < 1?) 2.12 Application: Evidence from Public Goods Games The prediction of the self-regarding model that all players choose to contribute nothing (e = 0) is consistently contradicted by the experimental evidence. The evidence we have comes from people playing one-shot (single-period) games and from people playing repeated games with as few as 5 rounds and as many as 50 rounds.13 In one-shot games, contributions average about half of the endowment, while in repeated games contributions begin around half and then decline so that a majority of players contribute nothing in the final round of a ten-round game. Researchers have interpreted the decline in the first half as a reflection of people getting disappointed about the expectations they had that other people would contribute more, along with the desire people have to punish low contributors (or at least not to be taken advantage of) in a situation in which one person can punish a low contributor only by reducing their own contributions. In this interpretation it is the reciprocity motives of the higher contributing subjects, disappointed or angry about their free-riding fellow subjects that explains why cooperation unravels. So the decline in contributions becomes a vicious circle: only by reducing how much they contribute can people punish others, but in so doing other people might want to punish them for their low contributions by contributing yet less again. Figure 2.14: Public Goods Game with Punishment. Average contributions over periods 1 to 10 decrease without punishment. Over periods 11 to 20, subjects can be punished by their peers and average contributions are higher on average than in the first 10 rounds. Source: Fehr and Gächter (2000a). The vertical axis is the average contribution each round. The horizontal axis is the period. At period 11 the subjects are given the opportunity punish each other. There are three treatments in this Public Goods Game experiment. This figure portrays the behavior in the "Strangers" treatment where players are randomly re-matched each round, but could have some players re-enter the group. The two other treatments, which show similar results, are "Partners" where players are in the same group for all the rounds; and "Perfect Strangers" where players are re-matched, but no player will encounter any other more than once during the experiment. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES The Public Goods with Punishment game The idea of that the decline in contributions is due to the fact that in the standard game contributing less is the only way to punish low contributors is supported by an ingenious experiment. This has the same public goods structure but with what turned out to be a major difference: after subjects contributed, their contributions were then made public to all the group members, and members then had the opportunity to punish others in the group. Subjects could punish them by imposing a cost, therefore, reducing the defectors total payoff. In order to impose this cost, however, the Punisher also had to suffer a cost themselves. The change in the rules of the game – adding the punishment option – represents a change in the institutions governing contributions to the public good. In the language of experiments the new rules are termed a new different treatment. So the standard game is one treatment and the game with punishment is a second treatment. In the experiment subjects engaged in extensive punishment of low contributors. At the start of the game people contributed over half of the endowment and then, apparently in response to punishment of low contributors, they contributed more over the course of the game. The change in institutions modeled by adding the punishment option altered the result dramatically. To see if subjects’ willingness to punish could be based on the expectation that they would benefit in subsequent rounds of the game, a slightly different experiment was tried. The researchers adopted what they called a “perfect strangers” treatment: after each round of the ten-round experiment the groups were re-shuffled, so that no player ever encountered any other player more than once. The "perfect strangers treatment" turned the experiment into a series of one-shot games. Since every player would only encounter every other player once, if lowcontributors responded to punishment by contributing more in subsequent rounds, they would raise the payoffs of others but not the punisher (who would never again be in the same group with the target of her punishment). In this way, punishment itself became a public good. This is because a punisher incurs a cost, yet the benefits of punishing a low contributor and getting them to increase their contributions are non-rival and non-excludable. Even in the perfect stranger treatment subjects avidly punished low contributors. Further evidence comes from the fact that people punish low contributors even in the last round of the game when punishment cannot be motivated by the expectation that the punisher will benefit from their targets improved behavior in the future. There is no future (the game ends after they punish). 105 106 MICROECONOMICS - DRAFT So reciprocal preferences – the pleasure of punishing a someone who is violating a social norm – are most likely involved.14 Culture matters When low contributors who are punished why do they subsequently contribute more? You may think that the answer is obvious: they contribute more to avoid the future reduced payoffs that being punished imposes on them. But there must be something else going on. In two similar experiments – one in the laboratory in the U.S. and one in the field among farmers in Zimbabwe “punishment” was not in reduced payoffs, it was just a purely verbal expression of displeasure (e.g "selfish guy") by a fellow subject.15 But the targets of purely verbal punishment contributed more in subsequent rounds. This occurs most likely because in many societies there is a norm that people should contribute to the public good, and when a person is criticized for violating it, they feel shame and try to make amends.16 Culture affects experimental play in other ways. The anthropologist Jean Ensminger conducted public goods experiments with the Orma, a herding people in Kenya. Members of the Orma regularly voluntarily contribute their labor to producing some public good – for example, the repair of a road – a system they call in Swahili, "Harambee." Families that have more cattle – more wealth – are expected to contribute more to the project. When Ensminger explained the Public Goods Game to the Orma participants, they promptly called it the “Harambee game.” Those with more cattle contributed more in the experiment, just as would have been the case in a real "Harambee." When the Orma subjects played the Ultimatum Game, however, they did not compare it to "Harambee" and wealth did not predict any aspect of their experimental play.17 This difference in how the wealthy played the Public Goods game and the Ultimatum Game probably would not have surfaced among the farmers you have already met from, Palanpur, Illinois or West Bengal. 2.13 Social preferences are not "Irrational" People sometimes think of other-regarding and ethical preferences as something special – different from the taste for ice cream, for example – and requiring a model different from the preferences, beliefs, and constraints approach. But the desire to contribute, to punish those who do not contribute, and otherwise to act on the basis of social preferences, like the desire to consume conventional goods and services, can be represented by preferences that conform to standard definitions of rationality. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 107 What we know from experiments is that whether its ice cream or contributions to the public good, people respond to trade-offs, taking account of the costs of and how much they value the activity in question: the higher the cost of helping others, the less its frequency. In other words, other-regarding preferences are consistent with rationality, namely consistency (transitivity) and completeness. Researchers tested the rationality of seemingly altruistic choices by asking 176 subjects to play a version of what is called the Dictator Game.18 In what is called the “Dictator Game", one player (the Dictator), Alice, is given a sum of money by the experimenter, and asked to transfer whatever proportion of the money that she wishes to an other (anonymous) subject, Bob. Alice is told that that for every dollar that Bob receives from her, she will have to pay p dollars. So p is the price of altruism:how much she has to pay for every dollar that Bob gets. After Alice makes her decision, the money is transferred, and the game is over. In this experiment, 75 percent of the Dictators gave away some money, demonstrating altruistic preferences. The average amount given away was a quarter of the endowment when the price p = 1 (a dollar for-dollar transfer).19 Moreover, the higher the price of generosity, the less money was transferred. For instance, when each dollar transferred to Bob cost Alice two dollars ( p = 2), only 14.1 percent of the endowment was given away on average, and when each dollar transferred cost four dollars, only 3.4 percent of the dictator’s endowment was transferred. The higher the price of altruism, the less did Alice "purchase." It may be, as the old saying goes, that "virtue is its own reward." But that does not mean that people will act virtuously no matter what the price. This finding is perfectly consistent with the fact that people respond to the price F AC T C H E C K In a Public Goods Game with Punishment experiment researchers found that the level of punishment that subjects inflicted on others was less when each dollar subtracted from the payoffs of the target cost more in foregone payoffs to the punisher.20 of virtuous behavior just as the preferences, beliefs, and constraints model predicts. Checkpoint 2.9: Dictator Game? Is the "Dictator Game" a game? Think about how we’ve defined games (check back in Chapter 1 to ensure you remember). 2.14 Application: The lab and the street Do people behave in the real world the way they do in experiments? The experimental evidence for reciprocity or related forms of other-regarding behavior would not be interesting if it did matched by similar behavior outside the lab. We therefore need to check whether laboratory evidence is externally valid, that is, consistent with behavior observed outside of the laboratory in similar circumstances to those found in the lab. External validity is particularly E XTERNAL VALIDITYResults of experiments or other scientific research that can be generalized to circumstances outside (external to) the laboratory or other setting in which the research was produced, are said to be externally valid. 108 MICROECONOMICS - DRAFT important for policy questions because policy-makers and governments need to know whether a policy will work outside of the controlled conditions of the laboratory. Generalizing directly from experiments to behavior in other contexts is often unwarranted. For example, in the Dictator Game typically more than 60 percent of the Dictators allocate a positive sum to the recipient, and the average given is about a fifth of the endowment.21 But we would be sadly mistaken if we predicted on the basis of this experimental result that 60 percent of people would spontaneously give money to an anonymous person passing them on the street, or that the same subjects would offer a fifth of the money in their wallet to a homeless person asking for help. Many researchers have tried to see whether behavior in lab experiments predicts behavior outside the lab. Along the coast of northeastern Brazil, for example, shrimpers catch shrimp in large plastic bucket-like contraptions. The shrimpers cut holes in the bottoms of the traps to allow the baby shrimp to escape, thereby preserving the stock of shrimp for future catches. The shrimpers face a real-world coordination problem: the expected income of each would be greatest if he were to cut smaller holes in his traps (increasing his own catch) while others cut larger holes in theirs (preserving future stocks). In Prisoners’ Dilemma terms, small trap holes are a form of defection that maximizes the individual’s material payoff irrespective of what others do (it is the dominant strategy). But a shrimper might resist the temptation to defect if he were both public spirited toward the other fishers and sufficiently patient to value the future opportunities that they all would lose were he to use traps with smaller holes. Ernst Fehr and Andreas Leibbrandt implemented both a Public Goods game and an experimental measure of impatience with the shrimpers. They found that the shrimpers with both greater patience and greater cooperativeness in the experimental game punched significantly larger holes in their traps, thereby protecting future stocks for the entire community.23 Additional evidence of external validity comes from a set of experiments and field studies with 49 groups of herders of the Bale Oromo people in Ethiopia, who were engaged in forest-commons management. Devesh Rustagi and his coauthors implemented public-goods experiments with a total of 679 herders, and also studied the success of the herders’ cooperative forest projects.24 The most common behavioral type in their experiments, constituting just over a third of the subjects, were reciprocators who responded to higher contributions by others by contributing more to the public good themselves. The F AC T C H E C K In an experimental game about trust and reciprocity played by groups of students and groups of chief executive officers of Costa Rican businesses, the businessmen were both more trusting of others and also reciprocated the generosity of their game partners to a far greater degree than did the students.22 Based on existing experimental evidence, students are not particularly other-regarding. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 109 authors found that groups with a larger number of reciprocators were more successful – they planted more new trees – than those with fewer reciprocators. This was in part because members of groups with more reciprocators spent significantly more time monitoring others’ use of the forest. As with the Brazilian shrimpers, differences in the fraction of reciprocators in a group were associated with substantial increases in trees planted or time spent monitoring others. 2.15 Application: A fine is a price Figure 2.15: A shrimping bucket with holes in it How might a policy-maker or CEO of a business make use of the fact that people care about what happens to others and they value behaving ethically? Think about a set of rules for compensating employees. The rules typically specify pay and provision for time off, sick days and the like. But problems arise with using purely material incentives to influence how people behave. Having noticed a suspicious bunching of sick call-ins on Mondays and Fridays, the Boston Fire Commissioner on December 1, 2001 ended the Department’s policy of unlimited paid sick days. Instead, the commissioner imposed a 15day sick day limit. The pay of firefighters exceeding that limit would be cut. The firefighters responded to the new incentives: those calling in sick on Christmas and New Year’s Day increased ten times over the previous year’s sick days. The Fire Commissioner retaliated by cancelling their holiday bonus checks. The firefighters were unimpressed: the next year they claimed 13,431 sick days; up from 6,432 the previous year.25 Many of the firefighters, apparently insulted by the new system, abused it, or abandoned their previous ethic of serving the public even when injured or not feeling well. In the language of the Ultimatum Game, they responded reciprocally to an offer they disliked by rejecting it. They were trying to punish the Commissioner at a cost to themselves. The Commissioner’s difficulties are far from exceptional. Consider the following experiment in Haifa, Israel.26 Parents everywhere are sometimes late in picking up their children at day care centers. • Treatment: At six randomly chosen day care centers, a fine was imposed for parents picking up their children late. • Control: In a control group of day care centers no fine was imposed. 110 MICROECONOMICS - DRAFT Fines Group with fine Control group 20 ● ● ● ● ● ● ● Late Arrivals ● ● ● 18 ● ● Figure 2.16: The effect of a fine for lateness in Haifa’s day care centers. Source: Gneezy and Rustichini (2000a). The fine was imposed in week 5 and retracted in week 17. ● 16 ● 14 ● 12 ● ● ● ● ● 10 ● ● ● ● 8 ● ● ● ● ● ● ● ● ● 1 2 3 4 ● ● ● ● 6 ● ● 5 ● ● ● 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Week Number Researchers expected parents to arrive on time because of the fine. But parents responded to the fine by arriving late more often: the fraction of parents picking up their kids late more than doubled. when the fine was taken away after 16 weeks, the parents continued to arrive late, showing no tendency to return to the status quo prior to the experiment. Over the entire 20 weeks of the experiment, there were no changes in the degree of lateness at the day care centers in the control group. The researchers reason that the fine was a contextual cue, unintentionally providing information about appropriate behavior. The effect was to convert lateness from the violation of a social norm or obligation that the parents were to respect, to a choice with a price that many were willing to pay. They titled their study “A Fine is a Price” and concluded that imposing a fine labeled the interaction as a market-like situation, one in which parents were more than willing to buy lateness for money. Revoking the fine did not restore the initial context. When monetary incentives undermine social preferences as they did among the Boston firefighters and Haifa parents, this is called crowding out. These two cases of crowding out are cautions that the use of monetary incentives may be inappropriate where the targets of the incentives are motivated by other regarding preferences. But they are not reasons to think that incentives are ineffective, as we will see in many examples to follow. We have no doubt that had the fine for lateness in Haifa been 500 New Israeli Shekels rather than 10 the parents would have found a way to pick up their kids on time. C ROWDING OUT is said to occur when monetary or other material incentives undermine other regarding or ethical preferences. 2.16 Complexity: diverse, versatile, and changeable people The experimental and observational evidence suggests an adequate understanding of preferences should recognize four aspects in human social PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES 111 behavior. • Diversity : people differ in their preferences . • Versatility : even a single person has a diversity of preferences, and which of these is salient for making a decision depends on the situation, for example, shopping as opposed to spending time with friends. • Changeability : people learn new preferences under the influence of their experiences. These three aspects of our preferences contribute to a fourth attribute of how human beings interact: • Complexity or "the whole is not the sum of its parts": the outcome of an interaction of many people cannot be deduced in any simple way from the characteristics of the individual people involved. Diversity What motivates people differs, both locally and across different cultures and across time. Using data from a wide range of experiments, researchers estimate that between 40 and 65 percent of people exhibit other-regarding preferences of some kind. The same studies suggest that between 20 and 35 D IVERSITY AND H ETEROGENEITY We use the term heterogeneous to describe a group of actors with different preferences or some other attributes, for example, wealth gender, nationality, first-mover status in game, and so on. percent of the subjects exhibit conventional self-regarding preferences.27 The authors of another study (in the U.S.) termed 29 percent of their experimental subjects as "ruthless competitors" (presumably resembling Economic Man) and 22 per cent as "saints."28 Versatility A common observation about human behavior made by psychologists is that the same person can act differently depending on the situation. As a result, we say that people are versatile: they change in response to what their situation seems to require of them, for example, being self regarding while shopping and other regarding with one’s neighbors. In the Ultimatum Game, Proposers often offer amounts which maximize their expected payoffs. But Responders rarely do. Researchers have also found this in experiments where subjects play both roles: Proposer at one stage in the experiment and Responder in another stage. The same person when in the role of Responder typically rejects positive offers if they appear to be unfair, even if they had made a similar low offer when in the role of Proposer. They therefore act as if they had reciprocal or inequality averse preferences as Responders, despite exhibiting self-regarding preferences when they are Proposers. The fact that in the role of proposers people are more like "ruthless competitors" while in the role of responder are more like "saints" is evidence of our versatility. V ERSATILITY How people act depends very much on the situation, resembling Homo economicus in some contexts (say, in business), but other-regarding social preferences in other contexts (say, around their family). Psychologists explain how a situation can frame a decision so as to suggest appropriate attitudes (or as economists would say, preferences) towards the possible actions an individual might make. We refer to this aspect of our behavior as versatility. 112 MICROECONOMICS - DRAFT Changeability Some preferences are part of our genetic makeup, having a taste for sweet and fatty foods, for example. But most preferences are learned rather than given by our genetic inheritance. Durable changes in an individual’s evalua- F AC T C H E C K In experimental games about dishonesty, people who grew up in Communist Party ruled East Germany are more likely to cheat than those who grew up in West Germany.29 tions of outcomes often take place as a result of experience. When this occurs we say that preferences are endogenous, meaning that they change as a result of influences such as where a person lives, how they make their living or the rules of the game that govern how they interact with others. Over a lifetime or even generations, migrants to a new country, or those moving from a rural to an urban area often adopt new preferences (for example concerning food tastes). The fact that preferences are learned may account for the fact that, as we saw from the experiments in small scale societies, people who hunt large animals tend be generous with the meat they acquire; and they seem to generalize these habits to other realms of life. Preferences are exogenous of they do not change or change only in response to influences that considered to be external. A consequence: Complexity In everyday language the word "complexity" refers to the state of being intricate or complicated. The term is used in quite a different way in the study of interactions of a large number of independent entities – whether particles or people. A key idea is that the results of these interactions for the system as a whole cannot be predicted in any simple way from even the most detailed knowledge of the interacting entities. The best example of complexity in the social sciences is Adam Smith’s invisible hand. What Smith suggested two and a half centuries ago, and modern economics has shown (as seen in Chapter 15) is that under some conditions uncoordinated interactions among entirely self-regarding total strangers through competition in markets among private property owners can (unwittingly) create an outcome that is better for all than many of the alternatives. The idea of complexity is often expressed the adage: the whole is different from the sum of the parts. The key here is not that the whole may be greater or less than the sum; it is that summing the parts is not the right way to calculate the whole. Averaging the components of some interacting system will not give what their interactions will actually add up to. The results of the interaction – called their emergent property – may be surprising given the nature of the interacting entities. Here are some examples of surprises (with which you are already familiar) in the properties that emerge from people with heterogeneous and versatile preferences interacting. E NDOGENOUS PREFERENCESPreferences are endogenous if they change as a result of influences such as where a person lives, how they make their living or the rules of the game that govern how they interact with others. PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES • Small differences in the distribution of types – the presence in a population of a few people willing to punish those who do not contribute in a Public Goods Game, for example – can have large effects on how everyone behaves, getting the self-regarding people to act as if they were cooperators. You have seen this in Figure 2.14. • Seemingly small differences in institutions can make large and surprising differences in outcomes. Why did adding the punishment option so radically change the outcomes in the Public Goods Game? We know that cooperation – contributing to the public good – unravels in in the absence of the punishment option. But the incentives to punish would seem identical to the incentives to contribute to the public good in the first place: everyone would like someone else to bear the cost of punishing the free riders. So not contributing and not punishing should be the dominant strategy in this game. But we now know that that is not what we observe. • While imposing a fine or other cost on socially undesirable behaviors may create socially desirable outcomes in certain circumstances such as getting people to stop using plastic grocery bags , a fine on parents arriving late to pick up their kids backfired. We saw that the nominal fine decreased parents’ willingness to pick up their children on time when the viewed the fine as a price to pay for additional day care: the fine changed what they viewed as socially acceptable behavior. • Letting a self-regarding player be the first mover in a Prisoners Dilemma game when she knows that the other player has strong reciprocity motives can avert the coordination failure resulting in mutual cooperation. Letting the Reciprocator be the first mover would have the opposite result: both players would defect, resulting in the Pareto inefficient outcome. You can confirm this by doing Checkpoint 28 Checkpoint 2.10: Sequential Prisoners’ Dilemma: Self-interest versus reciprocity For a sequential Prisoners’ Dilemma game where the first player is selfinterested and the second player is reciprocal draw a game tree in which the Nash equilibrium may be (Cooperate, Cooperate) and explain why could occurs. 2.17 Conclusion We have explored various social interactions represented as games and also as studied empirically using experiments, such as the Ultimatum game and the Public Goods Game. While self-regarding preferences are represented an essential and powerful motivator for human behavior, we have also found that 113 114 MICROECONOMICS - DRAFT people behave cooperatively – viewing it as the right thing to do – and they enjoy behaving cooperatively. People dislike unfair treatment and enjoy punishing those who violate norms of fairness or cooperation. The evidence that social preferences are common does not, however, suggest that people are irrational. Indeed, as we have seen, the experimental evidence suggests strongly that when individuals give to others their behavior conforms to the requirements of rational choice. People respond to the price of giving, giving more when it costs them less to benefit the people who receive their money. The importance of other-regarding preferences thus, does not challenge the assumption of rationality or the preferences, beliefs and constraints approach. However, it does suggest that for many applications we should take account of people’s concerns for others and for doing the right thing. Making connections Preferences, beliefs, and constraints: This framework for analyzing decisions will be used throughout the rest of the book. Risk and uncertainty: Many, maybe most, of the important decisions that people make are risky because the resulting outcome depends on contingencies the probability of which occurring the actor does not know. The rules of the game and coordination problems: Sequential rather than simultaneous play may result in a better outcome in an Assurance Game (or even a Prisoners Dilemma). The reason is that the first mover can help to coordinate play in the game. Another example: allowing players to punish low contributors in a public goods game dramatically changes the outcome. External effects and Pareto-inefficient Nash equilibria: The public goods game illustrates an extreme form of positive external effects (each person’s contribution benefits everyone equally). Evidence: Economists have recruited novel experimental evidence – from the laboratory and the field – to examine our theories about how people behave. Economists have used the evidence to modify and improve existing models and to develop entirely new models of how people behave. Heterogeneity: People differ in their preferences (self-regarding, other regarding) and in the advantages associated with their positions (first mover, second mover) PEOPLE: SELF-INTEREST AND SOCIAL PREFERENCES Important ideas preferences beliefs constraints rationality self-interest social preferences fairness altruism reciprocity spite endogenous (preference) exogenous (preference) institutions ad hoc external validity laboratory experiment field experiment endowment ultimatum game public goods game punishment group membership complexity inequality aversion diversity changeability learning Mathematical Notation Notation Definition x P p () E (), p̂ z e M n p a contingency probability of a contingency to occur a player’s payoff a player’s expected payoff individual endowment in Public Goods Game individual contribution in Public Goods Game return factor to contributions in Public Goods Game number of participants in Public Goods Game price of altruism in Dictator Game Note on super- and subscripts: i: an individual; Early: the strategy planting early; Late: the strategy planting late. Discussion questions See supplementary materials. Problems See supplementary materials. 115 3 Doing the best you can: Constrained optimization DOING ECONOMICS “What a useful thing a pocket-map is!” I remarked. “That’s another thing we’ve learned from your Nation,” replied Mein Herr, “mapmaking. But we’ve carried it much further than you." “What do you consider the largest map that would be really useful?” “About six inches to the mile.” “Only six inches! “ he exclaimed. “We very soon got to six yards to the mile. Then we tried a hundred yards to the mile. And then came the grandest idea of all! We actually made a map of the country, on the scale of a mile to the mile!” “Have you used it much?” I enquired. “It has never been spread out, yet,” “The farmers objected: they said it would cover the whole country and shut out the sunlight!" Lewis Carroll, Sylvie and Bruno Concluded, 1893 Lewis Carroll, the author of this dialogue (not to mention Alice in Wonderland) was also a mathematician and a philosopher. The point Carroll made about maps also goes for economic models. Maps are useful because they convey the necessary information, not because they are an exact representation of the territory, as the people from Mein Herr’s country discovered. Carroll’s point? The map is not the territory. A good model is not reality, but it’s a helpful guide. This chapter will enable you to: • See how the preferences, beliefs and constraints framework from Chapter 2 forms the basis for mathematical models of economic behavior. • Recognize how preferences – whether entirely self regarding or altruistic – can be represented both in mathematical form (a utility function) and graphical form (an indifference curve map). • Understand that constrained optimization is a method that economists use to explain the actions that people take; it is not a description of the thoughts or feelings making up individuals’ decision making processes (e.g.studied by a psychologists). • Explain how people are constrained – for example by limited time – and how these constraints give rise to opportunity costs and, along with our preferences, to trade offs. • Use the preferences, beliefs and constraints framework to analyze difficult choices concerning in policy-making, including how much of society’s resources should be devoted to the abatement of environmental damages. • Use the concepts of ordinal and cardinal utility explain how they differ and how cardinal utility provides a way to represent the societal cost of economic inequality. • Understand the shortcomings and limits as well as the insights of these models. What qualifies a map or a model as useful depends on what we need it for: six inches to the mile might be adequate for a map of hiking trails, but such a hiking map would not be much use to an airplane pilot. The same is true of economic models. Figure 3.1: The London underground transit system: the map represents, but is not the same as, the territory to which it corresponds. The map is a helpful model. 118 MICROECONOMICS - DRAFT Think of a model as a lens. An economic model is a way of focusing on what is important given the question that one wants to address without complicating the picture with things that do not matter for the question at hand. A key component of many economic models – those using the preferences, beliefs and constraints approach – is that we can understand the actions people take by assuming that they are doing the best they can under the circumstances that they are in. When implemented using mathematical reasoning, this process is called "constrained optimization, a process by which a person determines a course of action to accomplish a goal (reflecting the person’s preferences), given the information that the person has (beliefs) and the actions they may feasibly takes (a constraint). C ONSTRAINED OPTIMIZATION is the mathematical representation of a process by which a person determines a course of action in order to accomplish a goal (reflecting the person’s preferences), given the information that the person has (beliefs) and the actions they may feasibly takes (a constraint.) We illustrate a model and the process of constrained optimization by something that matters to all of us: Time, and how we use it. 3.1 Time: A scarce resource Benjamin Franklin (1706-1790) – the American politician and inventor – once said, "Time is money." Franklin was referring to the presence of trade-offs in how people choose to spend their limited time. His three-word sentence is therefore a constrained optimization model: people choose their daily actions to achieve their goals under the constraint of limited time. Spending an hour or minute on an activity provides us value of some kind: we enjoy the activity itself (e.g. eating) or the results of the activity (e.g. being paid a wage). But, since time is limited, choosing one activity also means we give up that time to do something else. We incur a cost of doing an activity because we forfeit the value of the next best thing we could have spent our time on instead: this is the opportunity cost of our time. Unless we have time to spare, and are wondering how we will fill up our day there is an opportunity cost to our use of time. As a result, we can model how we use our time as the result of our evaluating the benefits and costs (including opportunity costs) of pursuing one set of activities rather another. To do this we use constrained optimization. Before developing the concepts on which constrained optimization is based, let’s look at the kinds of facts that a model of time use should be able to explain. Figure 3.2 shows how men and women from the USA used their time each day during the year 2013. The largest time use is for the categories sleep, work (meaning for pay), leisure, and house work. Men and women differ typically in the hours they devote to paid work and house work and care work, often reflecting differing social norms about the kinds of activities that it is "appropriate" or "natural" for men and women to do. O PPORTUNITY C OST The opportunity cost of x in terms of y is the marginal rate of transformation: how much y a person must give up to get a unit more of x . D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 119 Figure 3.2: Daily time use of American men and women. These data – for hours in each activity measured on the horizontal axis for all adults – differ from data restricted to those with small children, or retired people, or students. 8 Hours 6 Gender Men 4 Women 2 pi ng ee k Sl W or d or ts Sp & e Le is ur Pa i k ew or k ou s e w or H D & ar rin ki lC Ea tin g C e ar ng so na op pi Pe r Sh ng 0 Source: Hofferth, Flood, and Sobek (2013). But these social norms also change, sometimes in ways that show that the differences in the distribution of work time between men and women are far from determined by "nature" but instead reflect changed economic conditions. During the second half of the 20th century in the rich countries the fraction of women doing paid work outside the home dramatically increased. While we do not have detailed information like that shown in Figure 3.2 for the mid-20th century on how men and women spent their time there almost certainly has been a decline in the amount of time doing housework. Part of the change in the distribution of women’s time between house work and work for pay is due to the availability at affordable prices of new technologies – household appliances – that reduced the amount of time required to clean house, wash clothes, and the carry out the other housework tasks. These include washers, refrigerators, and vacuum cleaners which in the U.S. became common from the late 1940s onward, and dryers, dishwashers and microwaves somewhat later. Evidence that these new technologies contributed to the change in the distribution of women’s work time comes from a comparison across countries of increases in the fraction of women working outside the home – called the labor force participation rate – and decreases in price of these labor saving household appliances (compared to other prices).1 The results are in Figure 3.3, which shows that in countries such as the U.S. where the prices of these appliances fell the most, women’s labor force participation rate rose the most. By contrast, in Germany where prices of household appliances fell the least, the increase in labor force participation was half as great as in the U.S. Other factors contributed, of course, most importantly the reduction in the F AC T C H E C K For a long historical view of why the washing machine was a "miracle" have a look at this video by Hans Rosling: <LINK HERE> - DRAFT MICROECONOMICS Change in Female Labor Force Participation 120 15 Figure 3.3: The relative price of home appliances and the female and male labor force participation rates. The vertical axis represents an index that records the change in the fraction of adult women working outside the home, termed the female labor force participation rate (FLFP), as well as the change in the home appliance price index (HAPI) on the horizontal index. The figures shows that as the price of labor saving household appliances decreases, the female labor force participation rate increases. Notice that a bigger price decrease would be shown by a larger negative change (further to the left on the x-axis) so the US, Denmark and the Netherlands had big decreases in the prices of home appliances and a big increase in the female labor force participation rate. Household appliances – like TVs – that did not reduce the amount of time necessary to perform housework tasks are excluded. Source: de V. Cavalcanti and Tavares (2008). Netherlands 14 13 12 11 US Denmark 10 UK 9 Luxembourg 8 Italy Belgium Ireland France 7 Germany 6 5 −0.30 −0.25 −0.20 −0.15 −0.10 Changes in the Home Appliance Price Index number of children born per woman. But the fall in the prices of appliance, the study concluded, was of approximately equal importance. It appears that economic changes – the new household appliances and their falling prices – changed how women spent their time – more working outside the home . This in turn may have been a both a result and a cause for the changing social norms about "women’s work" and the decreased adherence to the ideal of a family with a husband income earner and a wife raising (many) children and taking care of the home. This is an example of how preferences – for example, these norms – change as economic conditions – the prices of home appliances – change. We begin with these examples because methods of constrained optimization – the preferences, beliefs, and constraints framework – provide a way of posing and in some cases answering questions like: Why do men and women spend the time they do on the various activities shown? Or why did work hours fall so dramatically in some countries over the 20th century? We begin with with preferences, before turning to constraints later in the chapter. Because we are not considering strategic interactions or other situations in which the relevant facts are not known, we do not treat beliefs until the next chapter. Checkpoint 3.1: Labor-saving household appliances and women’s labor force participation Imagine a conversation around the year 1970 between a husband and wife who D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 121 just learned that a very effective clothes washing machine is available at a low price. How might the conversation have led to the woman taking up paid work outside the home? 3.2 Utility functions and preferences In Chapter 1, we represented preferences – our evaluations of the outcomes our actions may bring about – as payoffs that is, numbers indicating how much the decision maker values each of the possible outcomes. We discussed, as an illustration, the choice of whether to take an umbrella or not, with a decision [Don’t take the umbrella, It rains] resulting in a payoff of 3. The payoff to [Take the umbrella, It rains] was 15, meaning that if it rains the person valued having the umbrella by 5 times as much as not having it. In that example we simplified things by limiting the actions and the outcomes to just a few, for example, it either rained or it did not. The simplification allowed us to focus on 2 by 2 payoff matrices with just four possible outcomes. But most of the economic interactions that we study are not that simple: we can contribute any amount to the public good (not just $10 or nothing), the farmers in Palanpur have the choice to plant a little bit earlier, or much earlier, and so on. Or, to return to the question of time: how we divide up our day among the activities in Figure 3.2 could be measured in variations of minutes devoted to each of the nine activities, giving us trillions of "outcomes" to choose amongst. We need a way of representing preferences when there are a great many outcomes, without expanding our payoff matrices to the unusable size of the 1:1 maps in the Lewis Carroll fable at the beginning of the chapter. Why we use utility functions to represent preferences To do this we use a utility function, a mathematical expression that translates the full range of possible outcomes into a person’s valuation of the outcome – her payoffs. The word "utility" (in ordinary language, "usefulness") is used to mean the same thing as "payoff." It is a number assigned to a particular outcome bundle that has the property that when choosing between alternative bundles, a person will select the one with the highest (utility) number. Both "utility" and "payoff" sound like some monetary or other amount of something you take home as the outcome of a game. But in economics utilities, like payoffs are not something you get or even experience. You don’t take them home, they are nothing more than numbers that indicate the course of action you will take. R E M I N D E R : P R E F E R E N C E S represent the favorable (positive) or unfavorable (negative) feelings that could lead a person to choose one outcome over another. Included are tastes (food likes and dislikes, for example), habits (or even addictions), emotions (such as anger and disgust) often associated with visceral reactions (such as nausea or an elevated heart rate), social norms (for example, those that induce people to prefer to be honest or fair), and psychological tendencies (for aggression, extroversion, and the like). Do not think about a preference or the number a utility function assigns to some bundle as "how much a person likes" the bundle. U TILITY FUNCTION A utility function is an assignment of a number u(x, y), to every outcome bundle (x, y) representing a person’s valuation of that bundle. This means that if given the choice between two bundles (x, y) and (x0 , y0 ), the individual will choose the first if u(x, y) > u(x0 , y0 ). 122 MICROECONOMICS - DRAFT For simplicity, we call this number "how much the person values the outcome" but the utility function tells us nothing about why the bundle has a higher number. It could be any of the reasons for the collection of pro or con evaluations that make up our preferences for some bundle, ranging from food tastes to addictions to ethical norms. What the function allows us to do is to take account of more complex outcomes than "Don’t take the umbrella" and "It rains." The decision maker, as before, will choose the actions the she believes will result in the highest utility outcome. Suppose that our decision-maker, Annette, an Uber driver, is deciding how much time to work, x, and what fraction of the resulting income to spend on food, y. The utility function then assigns a number – the level of utility – to each possible combination of x and y, say, work for 4 hours and 15 0 00 M - C H E C K We read x as ’x prime" and x as "x double prime". We usually denote a bundle other than (x, y) as (x0 , y0 ) to indicate a different composition of the underlying x and y. minutes and spend 35 percent of the resulting pay on food. Any other combination, say, work four hours and spend 40 percent of the resulting income on food, will be assigned another number, representing Annette’s valuation of that particular outcome. This assignment of numbers is a utility function, u(x, y): for every outcome (x, y) the value of the utility function is the number representing a person’s valuation of the outcome. Then if we know what combinations of x and y are available to Annette based on the relevant constraints, then we can predict the choice Annette will make, namely the combination with the highest utility. What do the utility numbers measure? We measure how much a person values various outcomes in two ways, either: • by indicating how valuable each is on some absolute scale, or • by simply ranking them in order. If Annette compares two bundles (or outcomes), namely (x, y) and (x0 , y0 ) with u(x, y) = 3 and u(x0 , y0 ) = 9 there are two statements we could make about Annette, one much more informative than the other: • Annette values (x0 , y0 ) three times as much as (x, y) and • Annette values (x0 , y0 ) more than (x, y) In the first case above, utility is a number indicating by how much Annette prefers (x0 , y0 ) to (x, y). Utility is therefore termed a cardinal measure (cardinality in mathematics refers to the size of something). In Chapter 2 we represented people’s preferences by the payoffs associated with particular outcome bundle of games like (x0 , y0 ) or (x, y). When we defined the expected payoffs to some course of action we added up the payoffs of each possible C ONSISTENCY Consistency (or transitivity) requires that when considering three bundles (x, y), (x0 , y0 ), and (x00 , y00 ), if (x, y) is preferred to (x0 , y0 ) and (x0 , y0 ) is preferred to (x00 , y00 ), then (x00 , y00 ) cannot be preferred to (x, y). Consistent preferences can never lead someone to make contradictory choices. C OMPLETENESSCompleteness requires that all possible outcomes can be ranked. For any two bundles (x, y) and (x0 , y0 ) either the person prefers (x, y) to (x0 , y0 )) or the person prefers (x0 , y0 ) to (x, y)) or the person is indifferent between (x0 , y0 ) and (x, y)). D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 123 outcome (weighting them by the probability of each outcome occurring). Doing this required that utility is a measure of size. The numbers representing payoffs in Chapter 2 are cardinal utilities In the second case the utility function gives us an ordering of better-worse for the pair of outcomes. When the utility function is measured in this way, we say that Annette has ordinal preferences or that utility is ordinally measured. Ordinal utility says nothing about how much better the preferred outcome. Instead of assigning numbers to the outcomes, in the case of ordinal utility, it would be clearer if we just assigned ranks, like instead of 1,2,3,4 and so on, we used 1st, 2nd, 3rd, 4th (and in cases of indifference: for example, tied for 7th). In the cartoon figure about the O RDINAL PREFERENCES Ordinal preference rank outcomes: e.g. (x, y) ( x 0 , y0 ) (x00 , y00 ), without specifying how much (x, y) is preferred to (x0 , y0 ) or (x0 , y0 ) is preferred to (x00 , y00 ). The assignment of numerical utilities representing ordinal preferences is meaningful only to express the ordering: u(x, y) > u(x0 , y0 ) implies only that the first bundle is preferred to the second but not by how much. Planting in Palanpur game (Figure 1.2), we listed the four possible outcomes as "Best, Good, Bad" and "Worst": this is an example of ordinal utilities. There is no way that we can say that the top ranked bundle is twice as good as the second-ranked bundle or ten times as good as the tenth-ranked bundle. Nor could we add up the ranks, saying, for example, that getting your second ranked bundle and your third ranked bundle with equal probability is as good as getting your first and fourth ranked bundle with equal probability. None of these statements make any sense. This is why when dealing with decisions involving risk, we used a cardinal measure. So, when we introduced the Palanpur farmers’ uncertainty about when the C ARDINAL PREFERENCE A cardinal utility function assigns a number to each outcome, with the property that the ratio of the numbers assigned to alternative bundles expresses the relative degree of the preference for the alternative bundles. For example, with a cardinal utility function, u(x, y) = 10u(x0 , y0 ) = 5u(x00 , y00 ) means that (x, y) is preferred ten times as much as (x0 , y0 ) which is preferred five times as much as (x00 , y00 ), and that (x, y) is preferred fifty times as much as (x00 , y00 ). other farmer or farmers would plant their crops, we needed to think about expected payoffs, which requires adding up the values that each farmer attaches to an outcome. Because you cannot add up ordinal measures, we gave the payoffs numeric values (the numbers in the payoff matrix) representing cardinal utility. For some questions in economics the ordinal – better or worse – meaning of utility is all we need to understand and predict the actions that people will take. But in many situations, those involving risk and uncertainty, as we have just seen, or in evaluating the effects of differing rules of the game – policies to ensure competition in markets or concerning fairness, for example M - C H E C K For simplicity, we generally restrict our analysis to outcomes that can be described in terms of two variables x and y, though it is straightforward to generalize this model to outcomes described by more than two variables. The actor therefore makes choices among "bundles" that combine different amounts of x and y. – addressed in Section 3.13, the cardinal measure is required. Checkpoint 3.2: Utility and payoffs Give examples of preferences that might lead people to act in ways that they would regret. 3.3 Indifference curves: Graphing preferences Indifference curves are a useful way to visualize a person’s preferences. Let’s illustrate the concept of an indifference curve by Annette, who is choosing among differing amounts of kilograms of coffee (x) and gigabytes of data (y). I NDIFFERENCE CURVE The points making up an individual’s indifference curve are bundles – indicated by (x, y), (x0 , y0 ) and so on – among which the person is indifferent, so that u(x, y) = u(x0 , y0 ) and so on.This means that all of the bundles indicated by points making up an indifference curve are equally valued by the person. 124 MICROECONOMICS - DRAFT 10 10 9 a 8 Gigabytes of data, y Gigabytes of data, y 9 7 6 5 b 4 3 c 2 1 a 8 uA > 4 Better than uA4 7 6 5 b 4 3 uA < 4 Worse than uA4 2 c uA4 = 4 1 0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 Kilograms of coffee, x (a) Consumption Bundles Every point given by the coordinates (x, y) in Figure 3.4a is a pair of the quantities of the two goods, called a bundle. Points a, b, and c therefore represent three bundles of differing amounts of coffee and data. Suppose that Annette ranks the points a, b, and c equally – she is indifferent among the three bundles – then these three points lie on the same indifference curve, as shown in Figure 3.4 b. Her indifference curve represents the combinations of bundles among which she is indifferent. This means that for either bundle a – 8 gb of data and 2 kg of coffee – or bundle b – 4 gb of data and 4 kg of 3 4 5 6 7 8 9 10 Kilograms of coffee, x (b) An indifference curve Figure 3.4: One of Annette’s indifference curves: coffee and data. The dark green indifference curve uA1 represents all the combinations of x and y that provide Annette (A) with the same level of utility, 4. The blue area above and to the right of Annette’s indifference curve shows combinations of the amounts of coffee and data that provide her with utility greater than 4. The light green area beneath her indifference curve shows the bundles of x and y that she values at less than 4. She would therefore rather choose a combination of x and y on the indifference curve shown than any point to the left or below it. coffee– or bundle c – 2gb of data and 8 kg of coffee, u(2, 8) = u(4, 4) = u(8, 2) = 4. Figure 3.4 b shows the indifference curve made up of all bundles for which Annette’s utility is equal to 4. Her indifference curve is labeled by a u with a subscript which represents the level of utility that is the same for all points on that indifference curve. Annette prefers to consume more of both data and coffee, so she would like to be anywhere in the blue-shaded area where her utility would be greater than 4). She would rather not consume less of both data and coffee, so she would not like to be down to the area shaded in green where her utility would be less than 4. The single indifference curve shown in Figure 3.4 b divides the space of all possible bundles of x and y into three categories: bundles that are respectively better or worse than any of the bundles making up u4 and bundles that are equally valued with a utility of 4. To understand a decision-maker’s choice we proceed in steps: • Step 1: In this and the next section we use many such indifference curves to evaluate all of the possible outcome of the decision. • Step 2: In Section 3.6 we then limit the decision maker’s choices to those B UNDLE A bundle is a particular allocation, in the case of two goods given by (x, y). A bundle that results from the choice of one or more decision makers is call an outcome bundle. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N Gigabytes of data, y 10 9 e 8 a 7 6 d 5 uA5 b 4 125 Figure 3.5: An indifference map for kilograms of coffee, x, and gigabytes of data, y. The quantity of good x is on the horizontal axis and the quantity of good y is on the vertical axis. Three indifference curves are shown: uA3 , uA4 , and uA5 , where the rank of the utilities is uA5 > uA4 > uA3 . The constant level of utility for uA4 = 4. Points a, b, and c all lie on u2 and give Annette the same utility of 4. Point d would give Annette lower utility and point e would give Annette higher utility (because every bundle is associated with some utility number, we could draw indifference curves through those points, and through any point in the figure). 3 c 2 uA4 1 uA3 0 0 1 2 3 4 5 6 7 8 9 10 Kilograms of coffee, x that are feasible for (that is, choices that are actually open for the decision maker to take). • Step 3: Finally, use the evaluations in Step 1 to rank all of the feasible outcomes, showing us the one the decision-maker ranks the highest. To take Step 1, an individual’s utility function allows us to rank all of the outcome bundles – all combinations of x and y – that the decision maker considers. We do this using what is called a set of indifference curves (also termed an indifference map) as shown in Figure 3.5. Figure 3.5 shows three indifference curves, u3 , u4 , and u5 , part of Annette’s indifference map. Annette prefers more of both goods – that’s why they are called "goods." Therefore, indifference curves to the upper right, like u5 , are higher, (corresponding to the blue-shaded area in Figure 3.4). Indifference curves representing less preferred combinations, like u1 are to the lower left (corresponding to the green-shaded area in Figure 3.4). Of the three indifference curves plotted on the indifference map of Figure 3.5, uA 1 provides Annette with her lowest utility, whereas uA 3 provides Annette with her highest utility. A different person, one who valued coffee more than Annette would have a different indifference map. If you think of her indifference curves as a kind of contour map, Annette can be pictured standing somewhere on a mountain wanting to get to the top. She might, for example be in the lower left corner of the contour map of a hill shown in Figure 3.6 wanting to reach the 800 meter top of the hill. Her utility is the altitude where she is standing, say, at a point on the 720 meters above sea level contour. Her indifference curves are the numbered contour lines on a map of the mountain she is climbing, each indicating loca- I NDIFFERENCE MAP An indifference map is a set of indifference curves selected so as to illustrate some concept or result. For example to compare two bundles or to identify an outcome bundle that is the outcome of the decision-making process. 126 MICROECONOMICS - DRAFT tions on the mountain the same height above sea level. A map, as the quotation at the beginning of this chapter reminds us, is a representation of territory. The territory represented by Annette’s indifference map is her evaluation all possible outcomes she might experience. An indifference curve runs through every point in the (x, y) plane, but just like maps that could not possibly show every contour line, we can plot only a selected number of them in any case. Annette wants to climb as high as she can up the utility-mountain as possible, given whatever limitations she faces, including her own physical capacities and possibly impassible cliffs blocking her way. As Annette advances up the mountain, she crosses contour lines, moving from lower to higher indifference curves. She is engaging in a constrained optimization problem. Checkpoint 3.3: Maps, Points and Bundles Sketch your own version of the indifference map in Figure 3.5. Add two new points to your graph: a. A bundle, labeled f, where Annette holds the same amount of y as she does at point b, but Annette prefers bundle b to f. b. A bundle, labeled g, where Annette holds the same amount of y as she does at bundle b, but which Annette prefers to bundle b. c. Having manipulated the graphs and thought through the ideas of indifference curves, explain why the following is true: Consistency of preferences implies that indifference curves cannot cross. 3.4 Marginal utility and the marginal rate of substitution Indifference maps are used to summarize the values that an individual places on differing bundles of goods. But goods need not be things like Annette’s coffee or data. Goods can be anything a person values, such as free time. (Indifference curves, as we will show later in the chapter, can also summarize the preferences people have about "bads" such as environmental degradation, that, unlike goods, are things that people would prefer to avoid). To see this, we will move from the choice about coffee and data, and think instead about a new person, Keiko (KAY-i-ko), who is a student making a choice about the use of her time. As Keiko progresses through her studies (no doubt fueled by coffee and using data), she has two important priorities, which she thinks of as "Living" and "Learning." • Learning comprises all the aspects of her life as a student that contribute to her goals of becoming an educated person and becoming qualified for an interesting career. Figure 3.6: A contour map of a hill showing altitudes. Indifference curves are similar to contour lines, which are composed of all the points in the landscape which are at the same altitude. The lower left quarter of the contour map resembles the indifference map in Figure 3.5. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 127 • Living comprises everything else, including keeping up with friends, meeting new people, and taking care of herself. As there are only so many hours in a day, and because Learning takes time, Keiko faces what is called a trade-off between Learning and Living, the more she has of one the less she will have of the other. So she is facing another constrained optimization problem. We explain in this chapter’s last but one section that constrained maximization is not a description of the the mental and emotional processes by which we adopt one course of action over another. It is a research strategy that we use T RADE - OFF A trade-off is a situation in which having more of something desired ( a "good") requires having less of some other "good" or more of something that the actor would like to have less of (a "bad"). to understand what people do, not how they come to do it. But to illustrate the method we will suppose the Keiko consciously maximizes her utility function subject to her only-24-hours-in-the-day constraint, by comparing the utility associated with each of the combinations of Learning and Living that are open to her. (OK only a student in economics would actually do this!) Keiko is a systematic and quantitatively oriented person, and decides to measure her Learning quantitatively with a number. In calculating her Learning, she takes account of feedback from her teachers, such as grades (marks), but also evaluates this feedback in terms of her own estimation of how much she has learned, such as how much her study is improving her writing skills and general understanding. Keiko measures her the amount of Living by the hours she can spend not studying, x, and the amount of her Learning by her personal rating, y. Key to how the preferences, beliefs and constraints approach works is the fact that for most of the things that we may value, if we have little of it, we highly value having more of it, but the more of the thing we have, the less valuable D IMINISHING M ARGINAL U TILITY What is sometimes called the "Law of diminishing marginal utility" holds that the marginal utility of any thing that we value is less the more of it that we have. will be the next additional unit that we could have. This is called diminishing marginal utility, where the new idea here is "marginal." M-Note 3.1: The meaning of marginal The change in the value of a function – like utility, u(x, y) – when just one argument of the function x or y changes is a basic concept in calculus. The partial derivative of the function with respect to an argument – that is either ux (x, y) or uy (x, y) – is an approximation of the effect of a small change in the argument on the value of the function, holding constant the other arguments. If the decision-maker increases her consumption of x by a small amount Dx, then her utility is u(x + Dx, y) ⇡ u(x, y) + ux (x, y)Dx. The marginal effect on utility of a change in some element in the bundle a person is consuming (x) is calculated as the size of the change in u relative to the size of some small change in x with no other changes in the bundle. This is the size of the effect (Du) divided by the size of the cause (Dx ), so ux (x, y) = Du Dx where Dx is small. Conventionally this is expressed as the effect on u of a one unit change in x. If the marginal utility of any thing that we value positively is less, the more of it that we have – diminishing marginal utility – then this means that: F AC T C H E C K Diminishing marginal utility in economics is often based on the psychological principle of satiation of wants, which states that satisfying our wants is pleasurable, that our wants (for example hunger) are limited, when the resources allowing satisfaction of wants are limited we satisfy our most urgent wants first, and that the more satisfied is the want (by eating) the less pleasure do we derive from further satisfying the want. 128 MICROECONOMICS - DRAFT 6 1 Marginal utility of Living, ux g Utility, u(x, y) 5 i 4 3 f 2 1 0 f i g 0 0 2 4 6 8 Living, x 10 12 14 16 (a) Utility of Living holding Learning constant 0 2 4 6 8 10 12 14 16 Living, x (b) Marginal utility of living holding Learning constant at 3 • the first partial derivative of the utility function with respect to good x is positive, ux > 0, (because more x is better than less) and • the second partial derivative of the utility function with respect to x, uxx < 0 (because as x increases, utility is increasing (ux > 0,) but at a diminishing rate, therefore giving us diminishing marginal utility). Diminishing marginal utility A change of one variable – like Keiko’s Living – by one very small unit while holding constant everything else, including her Learning, is a marginal change, meaning the the change is very small and in only one variable. The Figure 3.7: Diminishing marginal utility. In panel a, utility is an increasing and concave function of Living, meaning that the curve it is positively sloped, but with a decreasing slope for higher levels of Living. The slope of the curve is the marginal utility, and this is shown in panel b. The points the in panel a, correspond to the same points in panel b. For example, the height of point f in panel a shows the level of utility when Keiko experiences 2 just hours of Living and the slope of a tangent to the curve at that point is the marginal utility of Living that point. The height of point f in panel b shows the value of that slope, that is the marginal marginal utility of increased Living when Keiko experiences 2 hours of Living. change in utility corresponding to a marginal change in x or y is called the marginal utility of x or y. Keiko’s marginal utility of Living which we denote as ux , like her utility itself, depends on how much Living and Learning she is currently experiencing. So we write ux as a function of x and y: ux (x, y). Similarly, Keiko’s marginal utility of Learning, uy (x, y) (or using the alternative notation Du(x,y) Dy , is how much her utility changes as she changes her Learning (y) by one unit, holding constant the amount of Living she does (x). To understand marginal utility, let us compare points f, i and g in Figure 3.8. By comparing these three points we can see how the marginal utility of Living (x) changes while holding Learning (y) constant. Keiko’s amount of Learning (y) is the same as she compares f, to i to g and the increase in her Living reA A sults in Keiko’s utility increasing from uA 1 to u3 to u3 . As her Living increases, however, each additional increase in x is associated with a smaller increase in utility, as reflected by her indifference curves getting flatter as she increases the amount of Living (that is, moving to the right along a horizontal line in the figure). This shows that the marginal utility of Living is decreasing as she gets more Living (increased x). Figure 3.7 shows just a slice of Keiko’s preferences, namely how they vary M-CHECK We also use the symbol for partial ∂ u(x,y) differentiation ∂ x to mean the marginal utility of x. When it is not necessary to be reminded of the other variables (held constant) that the marginal utility depends on, we eliminate the (x, y) and just use ∂∂ ux or ux . D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N Learning, y Figure 3.8: An indifference map portraying the choices between Living (x) and Learning y. The negative of the slope of the indifference curve is the marginal rate of substitution of Learning (y) for Living (x), mrs(x, y), capturing the trade-offs of Keiko’s preferences for the two goods. At f, Keiko has a high level of Learning (3) and little Living (2 hours) and she is willing to give up a lot of Learning to get more Living (her slope is steep at point f, therefore her marginal rate of substitution is large). At h, Keiko has a low level of Learning (0.82) and a lot of Living (14 hours) and she is willing to give up very little Learning to get more Living (her slope is relatively flat at point h, therefore her marginal rate of substitution is small). Keiko has a Cobb-Douglas utility function with u(x, y) = x0.3 y0.7 . Higher utility 4 g i f 3 uA3 2 uA2 Lower utility 1 h uA1 0 2 4 6 8 10 12 14 129 16 18 Living (hours), x with the level of Living she experiences, when the level of Learning she experiences is fixed at y = 3. We can study the full range of her preferences when the values of both goods varies by looking at her entire indifference map. The marginal rate of substitution This is shown in figure 3.8 where points f, i and g correspond to the same points in the previous figure. At point f, Keiko spends 14 hours studying and attending classes and has 2 hours left over for Living, with the result of a lot of Learning and not so much Living. At point h, Keiko spends 2 hours studying and attending classes and has 14 hours left over for Living, but her Learning is lower than at point f. Comparing between points f and h, we can see that Keiko sacrifices some of one good to get more of the other. Comparing f to h, Keiko gives up Learning to get more Living, but her utility remains the same. If we apply the same reasoning to very small differences in the quantity of the two goods, we can see that at point f the largest amount of Learning that 130 MICROECONOMICS - DRAFT Keiko would be willing to give up in order to get one more unit of Living is the negative of the slope of her indifference curve at that point (0.64 at point f), which is the marginal rate of substitution at that point or mrs(x, y). The fact that the indifference curve is steep at that point means she would be willing to give up a substantial amount of Learning to get a little more Living. This is because – as is clear from the previous figure – at point f she has an ample amount of Learning and not much Living, so her marginal utility of Living is high. Or, to put it a different way: the fact that at point f she has a lot of Learning means that the marginal utility of the Learning that she would give up to get some more Living is not very large. So the opportunity cost to Keiko of trading some Learning or some Living is low. The marginal rate of substitution is the maximum amount of y that Keiko can give up to get a small unit more of x without lowering her utility. The marginal rate of substitution is also the amount of y that Keiko would view as substitute for losing a small unit of x. The marginal rate of substitution should be read as "units of good y per unit of good x." We show in M-Note 3.2 that the marginal rate of substitution is equal to the ratio of the marginal utilities of the two goods: mrs(x, y) = ux (x, y) uy (x, y) (3.1) This is true because the amount of y that compensates Keiko for a small loss of x is the ratio of her marginal utility of x, which tells us how much she misses the x she has lost, to the marginal utility of y, which tells us how much she appreciates the compensating gain in y. Equation 3.1 also tells us something more about Figure 3.8. We can use the idea of marginal utilities to understand Keiko’s marginal rate of substitution at f, i and g in Figure 3.8. At point f on indifference curve uA 1: • The marginal utility of x is high because Keiko has very little x. • Conversely, the marginal utility of y is low because Keiko has a lot of y. • Therefore, her mrs(x, y) = uux is large and as you have already seen she is y willing to give up a lot of Learning to get a bit more Living. But at point g on uA 3 the opposite is true: • The marginal utility of x relative to the marginal utility of y is lower because Keiko has a lot of x than she had at point f. M ARGINAL R ATE OF S UBSTITUTION The marginal rate of substitution is the negative of the slope of the indifference curve. It is also the willingness to pay for a small increase in the amount x expressed as how much of y the person would be willing to give up for this. This is sometimes called the offer price. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 131 • Therefore, her mrs(x, y) = uux is smaller and she is willing to give up very y little Learning to get more Living. This is why the indifference curve is flatter at point g than at point f This shows that because of diminishing marginal utility, for a given amount of Learning, the more hours of Living Keiko has, the less amount of Learning she is willing to give up to get another unit of Living. The same reasoning shows (and Figure 3.8 confirms) that if Keiko’s preferences exhibit diminishing marginal utility for both x and y, her marginal rate of substitution of y for x declines starting at point f as we consider points on an indifference curve having more x and less y. M-Note 3.2: The mrs is the ratio of marginal utilities To derive the marginal rate of substitution using calculus, we use the method of total differentiation (covered in the Mathematical Appendix). First of all, along an indifference curve the amount of utility is a constant, u(x, y) = ū. So, to find the slope of the indifference curve we ask what changes in the quantities of x and y (one increasing the other decreasing) are consistent with u(x, y) not changing. This is what total differentiation tells us. The reason is that when we totally differentiate the utility function with respect to its arguments we express the change in Keiko’s utility as the sum of the changes due to changes in her consumption of each good. The total derivative of this equation is: du = ux (x, y)dx + uy (x, y)dy = d ū = 0 Since her utility is constant on an indifference curve by definition, the change in her utility is zero. Recall, too, that the derivative of a constant like ū is 0. We can now re-arrange equation 3.2 to find the mrs(x, y): Subtract uy (x, y)dy from both sides Divide by uy (x, y) and dx uy (x, y)dy = mrs(x, y) = ux (x, y)dx dy ux (x, y) = (3.2) dx uy (x, y) dy dx is equal to the ratio of ux (x,y) the marginal utilities of the goods, u (x,y) . But the negative of the slope of the indifference y As a result, the negative of the slope of the indifference curve curve is the marginal rate of substitution of y for x, so we have shown that the marginal rate of substitution is the ratio of the marginal utilities. The mrs has the dimensions of an amount of good y per unit of good x because the marginal utility of y has the dimensions utility per unit y, and the marginal utility of x has the dimensions utility per unit x. The mrs and the willingness to pay The marginal rate of substitution provides us with an essential piece of information. Imagine that Keiko had some bundle (x, y) and she were offered the following exchange – trade away some of her y in order to get more x. The mrs tells us the greatest amount of y that she would be willing to give up to get M C H E C K When considering two goods – things that people value positively, like data and coffee, or living and learning – the indifference curves are downward sloping. That is, they have a negative slope. The negative of the slope of an indifference curve is just its slope with the sign changed. 132 MICROECONOMICS - DRAFT one more unit of x in such a trade. This why we call the mrs the willingness to pay y to get more x. Why does the mrs tell us her maximum willingness to pay? The answer is that if in return for another unit of x she gave up an amount of y equal to the mrs she would be moving from one point on her indifference curve to another point on the same indifference curve. This is how we constructed the mrs. So she would be no better off after the trade than before. She would happily pay less than the mrs to get one more unit of x because this would increase her utility (put her on a higher indifference curve). But she would not pay more. This is why we call the mrs the maximum willingness to pay. Before going on to the constraints facing Keiko we will now show how what you have learned so far can be used with an explicit mathematical function. 3.5 Application: Homo economicus with Cobb-Douglas utility In Chapter 2, we saw that people may be some combination of preferences including self-regarding altruistic, fair-minded, reciprocal, spiteful, and so on. Representing these preferences mathematically requires knowledge of what Keiko values including: • How important to her are Learning and Living? • Is her own Living and Learning all she cares about, or does she value other people’s Living and Learning. In this section we study the preferences of a self-regarding Keiko: she does not care about the Living and Learning of others. We use what is called a Cobb-Douglas function utility function to illustrate how we can model the difference it makes what value she places on the two elements in her choice bundle. Here is a Cobb-Douglas utility function: u(x, y) = xa y(1 a) (3.3) The size of a , which is a positive number less than 1, is a kind of baseline measure of how much the individual values x independently of how much x and y she has. In M-Note 3.4 we show that if Annette is consuming the same amount of x and y then the maximum number of units of y that she would be willing to pay for one unit of x is 1 aa . So if a = 0.4 she her willingness to pay for a unit of x would be 0.4 0.6 or two-thirds of a unit of y. We also show in Chapter 7 that the fraction of a utility-maximizing consumer’s budget that will be spent on good x is a . The fraction spent on y will be 1 a. H I S TO RY The Cobb Douglas function is named after the economist and later U.S. Senator Paul Douglas and his then Amherst College colleague, mathematician and economist Charles Cobb, who jointly came up with the function in 1928 for an econometric study of the contributions of labor and capital goods to output in the U.S. economy. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N When a person’s preferences are described by a Cobb-Douglas utility function then as long as the Keiko has some of each good, x > 0 and y > 0, the following will be true: • her utility uCD (x, y) > 0 is positive, and • her utility increases as she consumes more of either good x or y, meaning that the marginal utility of both goods is positive. Because the marginal utilities for both goods is positive, Keiko will select a bundle with more of each over one with less of either if both bundles are available to her. Here is an example of a Cobb-Douglas utility function where a consumer, Annette from earlier, has a stronger preferences for y than for x because a = 0.4 and (1 a ) = 0.6. uCD (x, y) = x0.4 y0.6 (3.4) Let’s assume that x is kilograms of coffee and y is gigabytes of data as we did earlier. Because of the values of a and (1 a ), Annette has a stronger preference for data than for coffee because a = 0.4 < 0.6 = (1 a ). When a > 0 and (1 a ) > 0, Cobb-Douglas utility functions have the property that a bundle must include some of both goods to be assigned a positive utility, so we consider cases in which the person has x > 0, y > 0. M-Note 3.3: Cobb-Douglas Diminishing Marginal Utility How do we check that marginal utility is diminishing? Let us examine the marginal utility of Living in the Cobb-Douglas utility function. For the moment, we keep the function general with a : u(x, y) Utility Function = x a y1 a (3.5) To find the marginal utility of x we differentiate Equation 3.5 with respect to x: Marginal utility of x ux = ∂u ∂x = axa 1 1 a y (3.6) For 0 < a < 1, the marginal utility of x is positive, that is ux > 0. Why? x and y are both positive, as is the parameter a , as is the exponent 1 a . The exponent a 1 < 0, but this simply means that x can be read as being in the denominator of the marginal utility. For example, for a = 0.6, the marginal utility of x is: ux = 0.6 y0.4 y =a x0.4 x (3.7) You can see from Equation 3.7 that the larger is x the smaller will be the marginal utility of x. To confirm that the marginal utility of x is diminishing, we need to differentiate the marginal utility of x with respect to x. That is, we need to find the second derivative of the 2 utility function with respect to x, ∂∂ u2 x , that is, to partially differentiate Equation 3.6 with respect to x. 133 134 MICROECONOMICS - DRAFT ∂ u2 ∂ 2x Change in ux Because 0 < a < 1, a = 1 ) x (a (a )(a 1 < 0. Therefore, a (a 2) (1 a ) y <0 1) < 0. Therefore, the rate of change of the marginal utility with respect to x is negative (marginal utility is diminishing), or what is the same thing: utility increases at a decreasing rate as x increases. Checkpoint 3.4: Positive utility for x and y Consider Annette’s consumption of coffee and data as described by Equation 7.14. a. Sketch a map of three indifference curves for Annette based on her utility function. b. Confirm that any bundle with positive consumption of coffee and data (x > 0 and y> 0), is assigned a positive utility. c. How would you confirm whether consuming one more coffee or data increases Annette’s utility? d. If either (both) coffee and data increase her utility, at what rate does Annette’s utility change for changes in her consumption of coffee or data? (Hint: Mathematically, think through how you would find a "change in a change"). M-Note 3.4: Cobb-Douglas Coffee & Data We can derive the marginal rate of substitution for the general Cobb-Douglas utility function we defined earlier for coffee and data. Remember that along an indifference curve utility is a constant, such as ū > 0. When we find the "change" of a constant like ū, that change is zero, therefore du = 0 as in the set of equations below. uCD (x, y) = xa y ( 1 du = ux (x, y)dx + uy (x, y)dy = 0 a) = ū To find the marginal rate of substitution, we need to find the marginal utilities of x and y. Consequently, we differentiate the utility function with respect to x to find ux , the marginal utility of coffee, and with respect to y to find uy , the marginal utility of data. ux = axa uy = (1 1 (1 a ) y a ) xa y ( 1 (3.8) a) 1 (3.9) We substitute the marginal utilities from equation 3.8 and 3.9 into the definition of marginal rate of substitution, mrs(x, y) to find the formula for the marginal rate of substitution. mrs(x, y) = dy dx = = Factorize out x 1 and y 1 = ux (x, y) uy (x, y) (1 axa 1 y(1 a ) xa y ( 1 a) a) 1 axa x 1 y(1 a ) (1 a ) xa y ( 1 a)y 1 D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N a y(1 Remember that x 1 = 1x and y 1 1 = y and cancel the terms xxa and (1 y mrs(x, y) = a (1 y a) x a) a) 135 : (3.10) For example, equation 3.10 shows that if Annette is consuming the same number of gigabytes of data and kilograms of coffee (say, 5 each) she will evaluate them at ratio (1 aa ) . The preferences for each good (a and (1 a )) determines the ratio at which Annette is willing to trade data for coffee, together with the amount of coffee and data she is actually consuming. You can see that if Annette had a different level of current consumption of the two goods, say, more x and less y her mrs would be lower. y Checkpoint 3.5: Diminishing mrs as x falls Go back to Figure 3.5 and explain why the mrs is lower at point c than at point b, and lower at point d than at point b. Can you say if the mrs is lower at point a than at point d? Or at point e than at point a? Equation 3.10 may help you answer this question. 3.6 The feasible set of actions Keiko’s preferences and the resulting utility numbers she assigns to each bundle are a reflection of what she wants to achieve, what her goals are. But her preferences do not tell us what she can feasibly obtain. To understand the bundles that are feasible for her, we need to know how she obtains Learning from spending her time Studying. We suppose that Keiko sleeps 8 hours every night and she is not considering changing that. Her choice is what she will do with the 16 hours in the rest of the day . The relationship between the time Keiko spends studying and the amount of learning she achieves is given by a function that shows for the time (in hours) spent Studying (h), how much Learning (y) results, y = f (h). This is what economists call a production function – a mathematical description of the relationship between the quantity of inputs devoted to production on the one hand and the maximum quantity of output that the given amount of input allows. Production functions are more often used to study things other than success in coursework, that is outputs such as meals served, lines of code written, or bushels of corn harvested. Keiko’s production function is depicted in Figure 3.9 a. From it you can see that to obtain Learning a Keiko must spend hours (h) Studying. Up to a maximum of 16 hours she can increase her learning by studying more. But starting from studying just a few hours, doubling the amount of studying she does does not double her Learning. We can see this by comparing points e’, i’ and g’. Four hours of study (h = 4), gets Keiko y = 1.75 points of Learning, as P RODUCTION F UNCTION A production function is a mathematical description of the relationship between the quantity of inputs devoted to production on the one hand and the maximum quantity of output that the given amount of input allows. 136 MICROECONOMICS - DRAFT 4 4 yfʹ = 3.75 yf = 3.75 gʹ Studying to Learning production function y = f(h) yiʹ = 3 Learning, y Learning, y i yi = 3 iʹ yeʹ = 1.75 g eʹ Infeasible e ye = 1.75 Feasible set 1 1 Feasible frontier 0 2 4 6 8 10 12 14 16 0 2 (a) The production of Learning by Studying 6 8 10 12 14 16 (b) The feasible frontier of Living and Learning shown by point e’. But 8 hours gets her just 3 units of Learning, far from a proportional increase. This is because if she has just 4 hours she focuses on the really important key points, while if she has 8 hours she gets into the details, which add to her Learning, but not as much as the key ideas. Keiko’s learning production function illustrates an important common economic phenomenon: diminishing marginal productivity. The marginal productivity of hours studying is the effect of a small increase in studying time on the resulting Learning. As you can see from the fact that the production function in figure 3.9 is flatter for more hours of study, this marginal productivity of Studying hours is diminishing. This is similar to diminishing marginal utility. Just as the person satisfies her most pressing needs if she has very limited expenditures, but can turn to frills if she has more to spend, Keiko focuses on the essential points if her study time is limited but can turn to the examples and further illustration if she has more time to spend. Because Keiko has two ways to use her time – Studying or Living – and her waking hours are just 16 we know that: Hours of Living = 16 Hours 4 Living (hours), x = 16 − h Studying (hours), h Hours of Studying This makes it clear that • She has just one decision to make not two: if she chooses hours of Studying, that also determines her Hours of Living. Figure 3.9: Production of Learning by Studying and the feasible frontier of Living and Learning. Points e’, i’ and g’ on the production function show combinations of hours of Study and the maximum amount of Learning she could accomplish in that time. Point i’ for example shows that if she studies 8 hours she could attain learning equal to 3 (she could also attain less if she did spent the "studying" time texting with friends.) The amount of Living that she can have is her 16 hours minus the time she spends learning, i.e. x = 16 h, as shown in panel b. Panel b. shows the feasible frontier (dark green curve), which is the border of the feasible set (shaded in green). The feasible frontier is just a flipped version of her production function. Points beyond the feasible set (shaded in blue) are infeasible or infeasible given the number of ours in the day and her Learning production function. In this figure the equation for the feasible frontier is 1 2 given by: y = 4 64 x . D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 137 • Because more time living means less time studying this means that the opportunity cost of living more is some amount of learning less. To see what this opportunity cost is see Figure 3.9 Panel b, showing the feasible set of outcomes that Keiko might experience. The feasible frontier shown there is the mirror image of the production function in the panel a. The horizontal axis is no longer Studying hours but instead 16 minus Studying hours, which is the amount of Living she can have for each level of Studying she chooses. At e, Keiko can Study for 4 hours, which means she is Living for 12 hours at e and her Learning is 1.75. Or (point g) she could study for 12 hourw and have learning of 3.75. All of the points like e, g, i and the rest of the feasible frontier are choices that she could make. The feasible set is the area bounded by the feasible frontier and the x and y axes composed of all combinations of Living and Learning that she could experience. M-Note 3.5: A Living-Learning feasible frontier A mathematical expression for the feasible frontier is: Feasible frontier y = ȳ c(x) (3.11) The parameter ȳ is the maximum amount of y when x = 0, the y-intercept of the feasible frontier: if c(0) = 0 then y = ȳ. The term c(x) is the cost of x, that is, how many units of y (Learning) one must give up to get the value of x (Living) that she chooses. Suppose Keiko’s feasible frontier between Living, x, and Learning, y, is described by the relation: y=4 1 2 x 64 The negative of the slope of the feasible frontier, (3.12) dy dx = 1 32 x. Checkpoint 3.6 Redraw Figure 8.3 to show a new situation in which either: a. Keiko discovers that by changing her diet she can get by perfectly well on 7 hours of sleep. b. She transfers to a new university where it’s more difficult to get high grades. 3.7 The marginal rate of transformation and opportunity cost Turning to Figure 3.9 we can also contrast two points on the feasible frontier in, such as points a and b. At point a, Keiko spends 14 hours studying and attending classes and has 2 hours left over for Living, with the result of a lot of Learning and not so much Living. At point b, Keiko spends 2 hours studying F EASIBLE FRONTIER The feasible frontier is the border of the feasible set, showing for any value of x the maximum value of y that is feasible, meaning, that the decision-maker can obtain. 138 MICROECONOMICS Learning, y 4 - DRAFT Figure 3.10: Utility maximization: Living and Learning. Keiko’s feasible frontier for Living and Learning is shown in green. Three of her indifference curves are shown by uA1 , uA2 and uA3 in blue (uA3 > uA2 > uA1 ). She maximizes her utility at the point on her feasible frontier on the highest indifference curve, that is, at point b (where the two are tangents). At point b, she maximizes her utility where the marginal rate of substitution equals the marginal rate of transformation by choosing to spend 8 hours Living which gives her a subjective Learning score of 3. a b 3 mrs(x, y) = mrt(x, y) uA3 2 uA2 c uA1 1 Feasible frontier 0 2 4 6 8 10 12 14 16 18 Living (hours), x and attending classes and has 14 hours left over for Living, but her Learning is lower than at point a. The difference between the two points on the feasible frontier illustrates another trade-off that is central to Keiko’s choice: more living means less learning. And vice versa. If we apply the same reasoning to very small differences of the two goods, we can see that the opportunity cost in less learning that is required to get more living is the negative of the slope of her feasible frontier at that point, namely mrt (x, y). Dy Dx . This is termed the marginal rate of transformation or The marginal rate of transformation is the smallest amount of y that Keiko has to give up to get a small unit more of x. The mrt is therefore Keiko’s opportunity cost of x in terms of y or the minimum amount of y she has to sacrifice in order to get a small unit of x. The interpretation of the mrt as the opportunity cost of the x-good plays a major role in the reasoning in this book. Opp. cost of x = - Slope of feasible frontier = Dy = mrt (x, y) Dx The marginal rate of transformation should be read as "units of good y per unit H I S TO RY : F R E E L U N C H The 1975 collection of essays by Nobel Laureate Milton Friedman titled There’s No Such Thing as a Free Lunch: Essays on Public Policy popularized the idea that there is an opportunity cost to having more of anything that we value.2 M ARGINAL R ATE OF T RANSFORMATION The marginal rate of transformation is the negative of the slope of the feasible frontier. It measures the sacrifice of the y-good necessary in order to get more of the x-good. It is therefore the opportunity cost of the x good in terms of the y good. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N of good x." The term “transformation” is used because we think of a movement downwards and to the right along the feasible frontier as hypothetically transforming (giving up) the y-good into (having more of) the x-good. There is nothing actually being transformed. You can also interpret the negative of the slope of the feasible frontier as how much Learning you can get by giving up one unit of Living. The opportunity cost of the x-good in therms of the y-good differs depending how much of each good Keiko has. If her feasible frontier exhibits an increasing marginal rate of transformation as is shown in the figure, then her marginal rate of transformation of y for x increases as she moves along the feasible frontier toward having more x and less y. In this case, Keiko has to sacrifice more y for x the more x and the less y she has. As Figure 3.9 showed, between the y-intercept and point g: • She needs to give up relatively little Learning (0.25 points) to get four hours of Living. • Keiko’s feasible frontier is relatively flat. • Her marginal rate of transformation is therefore small. • Therefore the opportunity cost of Learning for Living is low. Between point e and the x-intercept, however, the opposite is true: • She must give up a large amount of Learning (1.75 points) to get an additional 4 hours of Living. • Her slope is steeper. • Her marginal rate of transformation is higher. • The opportunity cost of Learning for Living is greater. Keiko’s feasible frontier demonstrates an increasing marginal rate of transformation, which is to say increasing opportunity costs of Learning, moving from the y-intercept down the curve (left to right) to the x-intercept. This occurs because of diminishing marginal productivity of studying time in producing learning. M-Note 3.6: The Marginal Rate of Transformation Using Figure 3.9, we saw that the marginal rate of transformation was the negative of the slope of the feasible frontier. Suppose the feasible frontier is described by the equation: y=4 mrt = 1 2 x 64 dy 1 = x dx 32 139 140 MICROECONOMICS - DRAFT In this case the mrt increases with x, and exhibits the property of increasing marginal rate of transformation, or increasing opportunity cost. If you substitute the values we used earlier, for four hours, eight hours and 12 hours of living, we can evaluate the mrt (x, y) at each point: 1 • mrt (x = 4) = 32 (4) = 18 1 • mrt (x = 8) = 32 (8) = 28 = 14 1 • mrt (x = 12) = 32 (12) = 38 As x increases, the marginal rate of transformation increases, illustrating the idea of increasing opportunity cost. 3.8 Constrained utility maximization: The mrs = mrt rule From the feasible frontier we know that when maximizing her utility, the limited time in Keiko’s day creates a trade-off. By combining the insights of feasible frontiers and indifference curves – as in Figure 3.10 – we can understand how Keiko will manage this tradeoff • Constraints: She can choose some point on or within her feasible frontier given by her production function and the limits of her time. • Preferences: From among the points in her feasible set, she will prefer the outcome bundle with the highest utility, meaning on the highest indifference curve. To understand Keiko’s constrained utility-maximizing problem, we contrast points a, b, and c in Figure 3.10. An outcome bundle (x, y) is constrained utility-maximizing if there is no other point in the feasible set with a higher utility. Point a is on Keiko’s feasible frontier and lies on indifference curve uA 1 . But, a is not constrained utility-maximizing because Keiko could increase her utility by increasing her Living time and decreasing her Learning, by moving along the feasible frontier to the southeast. By similar reasoning point c cannot be the highest indifference curve she can reach. Keiko’s constrained utility-maximizing point is b in Figure 3.10, the point on the feasible frontier that is on the highest indifference curve. We label it b because it is the point where Keiko does the best she can. Figure 3.10 suggests a useful way to think about Keiko’s constrained utilitymaximization problem. In the figure, we see that the constrained utilitymaximizing bundle is the point where Keiko’s indifference curve is tangent to her feasible frontier. This means the indifference curve and the feasible frontier have the same slope at the constrained utility-maximizing point. The slopes of the indifference curve and the feasible frontier express tradeoffs between the two goods. This is the basis of what we call the mrs = mrt C ONSTRAINED UTILITY- MAXIMIZATION An outcome bundle (x, y) is constrained utilitymaximizing if there is no point in the feasible set that is on a higher indifference curve. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N rule. M-Note 3.7: Equating mrs to mrt to find the constrained maximum Suppose Keiko’s utility for Living (x) and Learning (y) is described by a Cobb-Douglas utility function with parameter a = 0.4 and (1 a ) = 0.6: u(x, y) = x0.4 y0.6 We find her marginal rate of substitution by finding the marginal utilities and substituting uA them into the equation mrs(x, y) = uxA y uAx mrs(x, y) = 0.4(xA ) = uAy = 0.6(x ) uAx uAy = 2y 3x 0.6 A 0.4 (yA )0.6 ( yA ) 0.4 (3.13) Suppose her feasible frontier is described by the equation: y=4 1 2 x 64 Keiko’s constrained maximum must be on her feasible frontier. We find her marginal rate of transformation by differentiating y with respect to x: dy 1 = mrt (x, y) = x dx 32 To find Keiko’s constrained maximum, we use the two expressions above for mrs and mrt , equating them to find a point on the feasible frontier consistent with the mrs = mrt rule: 2y 1 = x 3x 32 (3.14) Then multiplying through by 32 x: y= 3 2 x =4 64 1 2 x 64 x2 = 64 x=8 y=3 Keiko spends 8 hours on Living, studies 8 hours, and achieves a Learning level of 3. Doing the best you can: The mrs = mrt rule Summarizing the results so far, in Figure 3.10 1. The negative of the slope of the feasible frontier is the opportunity cost of getting a unit more more of the x good, in terms of the amount of the y good forgone. 2. The negative of the slope of an indifference curve is a measure of the person s willingness to pay for a little more of the x good in terms of how much of the y-good she would be willing to give up to get an additional unit of the x good. Using these two statements we can see why point a in in Figure 3.10 could 141 142 MICROECONOMICS - DRAFT not be the utility maximizing outcome bundle. The indifference curve is steeper than the feasible frontier, so the value of getting more living exceeds the associated opportunity cost (2 above is grater than 1) So she could do better by giving up some Learning in favor of more Living. REMINDER: The opposite is true at point c: the feasible frontier is steeper than the indif- • mrs, the marginal rate of substitution, is the negative of the slope of an indifference curve. ference curve, so the opportunity cost of having more Living falls short of the value of an additional unit of Living. So she definitely would not want to give up more Learning to get more Living. mrs AND mrt • mrt , the marginal rate of transformation, is the negative of the slope of the feasible frontier. In fact, it means the opposite. By giving up a unit of Living she would get a substantial increase in Learning (that is what the steep feasible frontier means). Giving up a unit of Living could be compensated by a modest increase in Learning (that is what the flatter indifference curve means). So the benefits of giving up some Living in return for more Learning outweigh the cost. So any point like a and c where the feasible frontier and the indif- ference curve intersect cannot be the constrained utility maximizing output bundle. This gives us the mrs mrt rule: The the utility maximizing output bundle is a point where Slope of feasible frontier = Slope of indifference curve which requires that: Marginal rate of transformation = mrt = mrs = Marginal rate of substitution Or, what is the same thing Opportunity cost of x = Willingness to pay for x The rule expresses a simple and true idea: if the opportunity cost of something is less than your willingness to pay you should choose more of it (if you can) and if the opportunity cost is greater than your willingness to pay, you should choose less of it (if you can). But there are cases in which the utilitymaximizing outcome bundle is not a tangency of the feasible frontier and an indifference curve: • It may be that an difference curve is steeper than the feasible frontier, but there is no way to get more of the x good. In this case the slope of feasible frontier does not measure the opportunity cost of getting more of the xgood; that is impossible (its cost is infinite). The utility maximizing outcome bundle at point b in Panel b of Figure 3.14 an example of a case – called a corner solution – there the mrt = mrs rule does not work. • We show in M-Note XX that there are conditions under which a bundle such that mrt = mrs can also be a minimum not a maximum. We provide an example of this in Chapter 6. T HE mrs = mrt RULE In many of the models that we consider in the remainder of this book, the constrained utility-maximizing outcome is a point on the feasible frontier at which an indifference curve representing the trade offs between the decision maker’s objectives is tangent to the feasible frontier representing the opportunity costs of having more of one good in terms of the amount of the other good foregone. This is the point where the marginal rate of substitution is equal to the marginal rate of transformation. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N Checkpoint 3.7: Changes in Keiko’s preferences Find Keiko’s constrained utility-maximizing level of Living and Learning when a = 0.6 and (1 a ) = 0.4, so that she values Living more than Learning. M-Note 3.8: When the mrs = mrt rule fails The rule can fail to identify the constrained utility maximum under two conditions: when the maximum is a corner solution (so the rule is not satisfied) and when the rule is satisfied at a minimum rather than a maximum. Positing a case with diminishing opportunity cost of obtaining one good in terms of the other good foregone will illustrate both cases Setup. Assume that a person’s utility varies with the amount of goods x and y: u(x, y) x+y = and the feacible amount of good y is a function of good x: y(x) = (1 x)2 (3.15) The rule may select a minimum, not a maximum.The marginal rate of substitution and marginal rate of transformation are: ux uy dy mrt (x, y) = dx Equyating the mrs and mrt 2(1 x) = x⇤ = y⇤ = mrs(x, y) = Using Equation 3.15 = 1 = 2(1 = 1 1 2 1 4 x) Note that using x⇤ and y⇤, the utility is u = 34 . Alternatively, we could set (x, y) = (1, 0), or (x, y) = (0, 1): both allocations are in the feasible set. In both cases, u = 1, which is higher than the one that we have reached using the condition mrs = mrt . The condition mrs = mrt will not give the utility maximum if the second order condition is violated: the second derivative of the utility function with respect to the variables must be negative. Let’s calculate it, replacing Equation 3.15 into the utility function: u d dx ✓ du = ux dx ◆ du = uxx dx = x + (1 = 1 = 2>0 2(1 x)2 x) The utility maximum may be a corner solution. In the example the utility maximums at both x = 1 and y = 1 are corner solutions (only one of the goods is consumed.) The rule may be inapplicable. Where either the indifference curves or the feasible frontier are not smooth but instead are kinked (are not differentiable), the derivatives on which the mrs and mrt are based will not exist at some points. Trade-offs between goods and bads In many situations it is easier to understand decisions in terms of a trade-off between a good and a bad rather than a trade-off between two goods. Recall that a bad is something that you would prefer to have less of, such as working 143 144 MICROECONOMICS - DRAFT uA3 Higher utility uA2 3 b 2 mrs(x, y) = mrt(x, y) 3 Feasible frontier b 2 a Lower utility 1 2 uA1 c a 0 uA2 4 c Learning, y Learning, y 4 uA3 uA1 4 6 8 10 12 14 1 16 18 0 2 Studying (hours = 16 − Living), h 4 (a) Indifference curves of Learning and Studying 8 10 12 14 16 18 (b) The utility-maximizing choices with bads harder than is comfortable or safe. For example, Keiko might think of her decision in terms of a trade-off between her time studying time that she does not enjoy, h = 16 6 Studying (hours = 16 − Living), h x, and her Learning, y. The more time Living the better for Keiko, therefore x is a good. The more time Studying the worse for Keiko, therefore h is a bad. But as before since x = 16 h, choosing (h, y) to maximize utility, u(16 h, y) is the same thing as choosing x to maximize utility u(x, y). These are just different ways of posing the same problem. Figure 3.11 shows Keiko’s indifference curves and feasible frontier plotted in terms of Study time, h and Learning y. Her indifference curves slope upward because an increase in Studying, h, lowers Keiko’s utility, and requires an increase in Learning, y to compensate in order to stay at the same level of utility. Utility increases as we move to the northwest and decreases as we move to the southeast in this plot. Similarly, Keiko’s feasible frontier slopes upward, because an increase in Study time, h, leads to more Learning, y. This is her "learning production function" introduced earlier. So the slope of the feasible frontier is the marginal Dy productivity of studying time or Dh and this is also the marginal rate of transformation of Study time into Learning. (In this case "transformation" actually describes the process underlying the feasible frontier). As was the case for tradeoffs between two goods, a bundle in the feasible set is the utility maximizing output bundle if there is no other feasible bundle Figure 3.11: The mrs = mrt rule:. Keiko’s problem of choosing (h, y) when h = 16 x = time Studying, is a bad. Studying time, h is plotted on the horizontal axis, and Keiko’s Learning, y is plotted on the vertical axis. Keiko’s feasible frontier is shown in green in the right-hand panel. Three of her indifference curves are shown by u1 , u2 and u3 in blue in both panels. The points a, b, and c are the same as in Figure 3.10. Keiko maximizes her utility at the point on her feasible frontier on the highest indifference curve, that is, at point b, choosing to spend 8 hours on Living and 8 hours on Learning. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 145 with greater utility. And this is the bundle for which the mrs = mrt rule holds, namely the point on the feasible frontier where the marginal rate of substitution equals the marginal rate of transformation (mrs(x, y) = mrt (x, y)). M-Note 3.9: The marginal utility of the bad The utility function for Studying (h) and Learning (y) is given by: uA (h, y) = (16 h)0.4 y0.6 (3.16) To find the marginal utility of the "bad," Studying, we need to partially differentiate Equation 3.16 with respect to h. Remember that when we partially differentiate we treat the other variable as a constant, so the term y0.6 will simply remain where it is. We only have to think about the h term. ∂ uA = uAh ∂h uAh = (0.4)( 1)(16 = 0.4 (16 h) | {z } | {z <0 >0 h)0.4 1 0.6 y 0.6 y0.6 < 0 } |{z} >0 The first term is negative whereas the second and third terms are positive. So the marginal utility of hours of study is negative. We call such a utility a disutility and will often talk about the disutility of work or the disutility of effort. Checkpoint 3.8: Understanding goods and bads Find Keiko’s constrained utility-maximizing level of Study time and Learning, using the mrs = mrt rule. 3.9 The price-offer curve, willingness to pay, and demand We often want to know how people respond to different options for exchange in the form of prices. We may be interested in knowing, for each price at which she can purchase any amount of the good she pleases, how much Keiko will purchase, namely the utility maximizing amount. This is Keiko’s individual demand curve. Remember that in explaining Keiko’s indifference curves we asked what is the maximum amount of Learning she would be willing to give up in exchange for more Living. The answer is given by her maximum willingness to pay, or what is the same thing her marginal rate of substitution of Learning for Living. We now ask almost the same question except that rather than giving up Learning to get more Living, Keiko is now giving up money – that is paying for a good according to its price. For each offered price she faces another constrained utility-optimization problem. The demand curve is constructed by a series of hypotheetical constrained optimization problems, one for each possible offered price. Each I NDIVIDUAL DEMAND CURVEAn individual demand curve (or demand function) indicates for each price that might hypothetically be offered at which a buyer can purchase any amount that they please, the quantity that an individual will purchase. 146 MICROECONOMICS - DRAFT y=m y=m Feasible (within the budget) Budget Constraint Money left over, y Money left over, y a Infeasible (outside the budget) yb b u3 u2 y=m−p⋅x c u1 Budget constraint, bc1 Kilograms of fish, x x= m p (a) The budget constraint price defines a feasible set; its boundary, the feasible frontier, defines the bundles of goods Keiko has access to. For each each of these feasible sets there is a bundle that maximizes her utility. This is a single point on her demand curve. Indifference curves tell us the utility number that Keiko assigns to each possible consumption bundle. Using this logic, her choice will be the point on the feasible frontier with the greatest utility, which will be the bundle in the feasible set that is on the highest indifference curve. This is a standard constrained utility maximization problem. xb x= Kilograms of fish, x m p (b) Utility-maximizing choice Figure 3.12: Budget constraint and utilitymaximizing choice for fish and money for other goods. The budget set is shaded in green and the budget constraint (feasible frontier) is the dark green line on the border of the budget set (feasible set). Consumption bundles (x, y) in the budget set and on the budget constraint can feasibly be obtained with the current budget (m) at the price, p, for kilograms of fish, x, Outside the budget constraint, in the shaded green area, the bundles of x and y cannot feasibly be obtained with the existing budget. Harriet maximizes her utility subject to her budget constraint bc1 . She maximizes her utility at b where her marginal rate of substitution, mrs(x, y) = uux , equals her marginal y rate of transformation or the price ratio of x to y, mrt (x, y) = p. The budget constraint and feasible utility-maximizing choices We shall use one particular kind of feasible frontier to think this through: the budget constraint. The budget constraint defines an amount of money m that a person has or has access to, through wealth and credit markets, which constitutes their budget to spend on goods and services. People can use their budget to spend on goods at prices that are given to them. Imagine that you want to buy the fish that Alfredo and Bob were trying to catch in Chapter 5 at a fish market. The price ( p) is measured in dollars per kilogram. Figure 3.12 a. shows the budget constraint for Harriet, someone deciding on how much fish to buy from Alfredo or Bob at price p. The budget set is shaded in green and the budget constraint (feasible frontier) is the dark green line on the border of the budget set (feasible set). Consumption bundles (x, y) in the budget set and on the budget constraint can feasibly be obtained with the current budget (m) at the price, p, for kilograms of fish, x, Outside the budget constraint, in the shaded green area, the bundles of x and y cannot feasibly be obtained with the existing budget. B UDGET CONSTRAINT A person’s budget constraint gives the bundles (x, y) that just exhausts some given budget at a set of market prices ( p) of the goods. The feasible set includes all purchases bundles that do not exhaust the budget, so the budget constraint is the feasible frontier. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 147 We know how to find the utility-maximizing bundle for a given feasible frontier – or the budget constraint – with given indifference curves: we apply the mrs = mrt rule finding the bundle where the marginal rate of substitution equals the marginal rate of transformation. We can combine these insights and calculate what the consumer’s utility-maximizing bundle will be for every potential price of the good given a fixed budget and when the other good, y, is money for other goods. Figure 3.12 b shows Harriet maximizing her utility subject to her budget constraint bc1 . To find her utility-maximizing choice, we must apply the mrs mrt rule to find where her marginal rate of substitution (her willingness to pay in money for kilograms of fish) equals her marginal rate of transformation, here the price for a kilogram of fish. At point a she consumes too little of x and too much of y (her marginal utility of money for other goods (y) is much lower than her marginal utility of kilograms of fish (x), or her mrs(x, y) is too high, and she would be better off if she consumed less y and more x. Conversely, at c, she consumes too little of y and too much of x (her marginal utility of x is much lower than her marginal utility of y, or her mrs(x, y) is too low, and she would be better off if she consumed less x and more y. She maximizes her utility at b where her marginal rate of substitution, mrs(x, y) = uux , equals her marginal y rate of transformation or the price ratio of x to y, mrt (x, y) = p. The demand curve: Utility-maximizing choices at difference prices With every change in price, the consumer’s budget constraint will pivot. The budget constraint will pivot upwards as a good’s price decreases, because a consumer can buy more of the good with the same budget. The opposite is true for price increases. As the price of a good increases, the same budget buys less of the good, pivoting the budget constraint inward. P RICE - OFFER C URVE The price-offer curve shows every utility-maximizing consumption bundle at each price of good x. It demonstrates the principle of demand by connecting every point where a consumer’s indifference curve is tangent to every possible budget constraint for a change in the price of x at given income m. We will use the price-offer curve in 4. With every pivot of the budget constraint, at the utility-maximizing point, the new budget constraint will be tangent to a new indifference curve which will be either higher if the price of the good decreases or lower if the price of the good increases. Because we can calculate the utility-maximizing consumption bundle for each possible price, we can find a curve that records every utility-maximizing consumption bundle for each price, connecting up points a, b, and c in the left panel of the figure. That curve is called the price-offer curve. Sometimes, for individual consumers, it is called the price-consumption curve because it indicates what the consumer will consume at different prices. Figure 3.13 maps three different utility-maximizing consumption bundles at three prices of x. With each price decrease, the budget constraint pivots outward from p1 to p2 to p3 . With each change in the price of x, the utilitymaximizing bundle – the point at which the marginal rate of substitution is M - C H E C K We can find the equation for the price-offer curve by using the equation for the budget line and combining it with the equation for the marginal rate of substitution. We do not derive it here as it is not required to understand the intuition of the demand curve. 148 MICROECONOMICS - DRAFT Money left over, y Offer Curve c y3 b y2 u3 p1 = 0.25 u2 a y1 p2 = 0.5 u1 p3 = 1 Price per kilogram of fish, p x1 = 6 p=2 p3 = 1 x2 = 9 x3 = 10.5 Demand curve aʹ bʹ p2 = 0.5 cʹ p1 = 0.25 x1 = 6 x2 = 9 x3 = 10.5 Quantity of fish in kilograms, x equal to the opportunity cost – changes. At p3 = 1, the bundle includes x = 6, at p2 = 0.5, the bundle includes x = 9, and at p1 = 0.25, the bundle includes x = 10.5. With each price change, there is a new bundle for both x and y. The different bundles suggest a price-quantity relationship between the quantity demanded of x and different prices of x. As the price of x decreases, the quantity demanded increases. In fact, we can take each price-quantity combination and map a demand curve to it. In the lower panel of Figure 3.13, we have taken each utility-maximizing consumption bundle from the different consumption bundles at each price and identified their coordinates on price-quantity axes. The price-quantity combinations provide a downwardsloping demand curve where quantity demanded, x, decreases as its price, p, Figure 3.13: Offer curve and demand curve for fish: The price of x in the top panel is in terms of the money Jane sacrifices to get more fish. Similarly, in the lower panel the amount of money Keiko must sacrifice to get more fish – the price per unit of fish – determines Keiko’s quantity of fish demanded along the demand curve. Points a, b, and c in the top panel correspond to points a’, b’ and c’ in the lower panel. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 149 increases. Measured horizontally from the vertical axis, it tells us the amount that can be sold to the consumer at each particular price. Measured vertically from the horizontal axis, it also tells us what is the consumer’s maximum willingness to pay for each amount on the horizontal axis. 3.10 Social preferences and utility maximization The preferences we have looked at so far have been entirely self regarding, depicting a person who is concerned with their choices among bundles that they alone will experience. But people often make choices where they are not the only person affected, where what they choose can benefit or harm E X A M P L E We demonstrate in Chapter 7 how to find the equation for the demand curve and how to see that a reduction in the price of fish has two effects. First, the lower price leads the person to buy less meat and more fish; this is the substitution effect. Second, the lower price also allows the person to buy more of everything if she chooses (fish, meat or whatever); this is called the income effect. someone else. Consider the Dictator Game that we mentioned in Chapter 2. In that game, a person, the Dictator has an endowment of money, y, that they can choose to split between themselves and another person in any way they choose. Imagine that Annette is the Dictator and she is able to choose whether or not to give to some amount to Ben. As a result, Annette must choose some split of of her endowment z = p A + p B , where p A is the amount in dollars that Annette keeps for herself and p B is the amount that she gives to Ben. As a result, we can re-arrange the equation to find the equation to the feasible frontier for y = 10 dollars: Feasible Dictator Allocations p B = 10 pA (3.17) Looking at Equation 3.17, we can see that the feasible frontier is a straight line with a slope of 1. This tells us that the feasible frontier slopes downward. Remember that the negative of the slope of the feasible frontier is the marginal rate of transformation: so a player in the Dictator Game who wishes to give $1 to someone else has an opportunity cost of $1 for doing so. If Annette is like Homo economicus, she is purely self-regarding. She sets p B = 0 and keeps R E M I N D E R A game is a mathematical representation of a strategic interaction, which means one in which players recognize that their payoffs depend on the actions taken by other players. So the so-called Dictator Game is not really a game at all, because the Dictator’s payoffs do not depend at all on anything that the other player does. everything or herself and therefore z = p A = 10. What happens when the Proposer is an altruist who believes in making an offer of more than zero to a partner? To see what happens in these cases, let us contrast two pairs of people: M - C H E C K Two things to remember when thinking about Equations 3.18 and 3.19. • Annette (A) is paired with Ben (B). Annette makes choices about how much money she gets and how much money Ben gets. • Chen (C) is paired with Diane (D). Chen makes choices about how much money he gets and how much money Diane gets. To think about the choices that Annette and Chen make, let us consider two different kinds of Cobb-Douglas utility functions that Annette and Chen might • The exponents in the Cobb Douglas utility function Equation 3.19 mean that if they both had the same payoff, then Chen would value increasing his own payoffs more than he would value increasing Diane’s. • Any number raised to a zero exponent is equal to 1, so because Annette does not value Ben’s payoffs at all (the exponent is zero) her utility is unaffected by the amount that he gets (her utility is simply how much she keeps for herself). 150 MICROECONOMICS - DRAFT 11 uA1 uA3 uA2 11 9 9 8 8 D's payoff (dollars), πD 10 B's payoff (dollars), πB 10 Feasible frontier 7 6 5 4 3 7 6 5 mrs = mrt 4 2 1 1 b 0 1 2 3 4 5 6 7 8 A's payoff (dollars), πA 9 10 bʹ 3 2 0 aʹ uC 1 0 11 0 1 2 uA (p A , p B ) = (p A )1 (p B )0 (3.18) uC (p C , p D ) = (p C )0.7 (p D )0.3 (3.19) Equations 3.18 and 3.19 represent the utility functions of two different people. Chen is other-regarding, he cares about Diane’s payoff as is indicated by the positive exponent on her payoff in his utility function, though not as much as he cares about his own (compare the two exponents). Annette is entirely selfregarding, placing a zero weight on Ben’s payoff and therefore her choice will depend entirely on the outcome that she experiences. We display indifference curves for Annette and Chen in Figure 3.14. The indifference curves in panel a are unusual: they are vertical because the only thing that Annette values is what is on the horizontal axis, namely, her payoff. Using the mrs = mrt rule, we find the constrained utility-maximizing point for each person where their highest indifference curve touches the feasible frontier. In this case, though, the feasible frontier is given by a straight line because it represents a split of money. The maximum amount of money that Annette or Chen can keep is $10 and they can offer splits in 1 cent increments between themselves and their partners. The vertical intercept corresponds to the instance in which they give all $10 to their partners. The horizontal intercept corresponds to the instance in which they keep all $10 to themselves. Chen has preferences such that he would like a 70%-30% split of the $10 (his 3 4 5 6 7 8 C's payoff (dollars), πC 9 10 11 (b) Chen is altruistic, offering a (7, 3) split have. Chen’s Utility Function uC 2 cʹ (a) Self-interested Annette offers a (10, 0) split Annette’s Utility Function uC 3 Figure 3.14: Utility maximization: Self-interested offer vs. altruistic offer. Annette offers a split to Ben of (10, 0), whereas Chen offers Diane a split of (7, 3). Annette’s indifference curves are vertical because she gives no weight in her utility function to Ben getting any money ((1 a ) = 0), therefore she gets $10 and Ben gets $0. Between Chen and Diane, Chen gives some weight to Diane getting money ((1 a ) = 0.3), therefore his indifference curves are shaped like indifference curves we’ve looked at previously and at his constrained utility maximum Chen gets $7 and Diane gets $3. Notice that if Chen gives any less or any more to Diane, then he would be on a lower indifference curve, such as at points b’ and a’ on uC1 . D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N a = 0.7) and his highest indifference curve is tangent to the feasible frontier at a (7, 3) split shown by point b’ in Figure 3.14 b. Annette has preferences such that she would like a 100%-0% split of the $10 (her a = 1, she places zero weight on Ben’s payoff) and her highest indifference curve touches the feasible frontier at b in Figure 3.14 a at a (10,0) split (she keeps all the money). We can interpret the slope of her indifference curves as her maximum willingness to pay in order to give Ben a small positive payoff, and the ask: how much of her own payoffs would she be willing to give up to transfer a penny to Ben? The answer is that there is no amount, however small, that would motivate her to do this. But what allocation does she choose? She chooses the highest utility that is within the feasible set. Her highest utility is where her vertical indifference curve uA 2 touches her highest feasible allocation to herself of $10. She keeps all the money. Her keeping all the money shouldn’t surprise us because she gives no weight to Ben’s payoff. In mathematics, a solution like this is called a corner solution. Notice that we couldn’t use our standard requirement for finding the constrained utility maximum of mrs = mrt . mrs(p A , p B ) was undefined because her indifference curves were vertical. But the principle of constrained utility maximization, that Annette would find the point in the feasible set with the highest utility, still applied to our problem and we found the solution. M-Note 3.10: The mrs for a self-regarding Dictator Why are Annette’s indifference curves vertical in Figure 3.14? To answer this question, we need to find her marginal rate of substitution. To find her mrs, we need the marginal utilities of the two arguments of her utility function: p A and p B the money payoffs that Annette and Ben respectively get. Marginal utility to Annette of Annette’s payoff: uAp A = ∂ uA ∂ pA 1 · ( p A ) (1 = 1) (p B )0 = 1 Marginal utility to Annette of Ben’s payoff: uAp B = ∂ uA ∂ pB = 0 · ( p A ) 1 ( p B ) (0 1) =0 Therefore Annette’s marginal rates of substitution is: mrs(p A , p B ) = = up A up B 1 = undefined 0 (3.20) Now, the result of Equation 3.20 should not surprise us because the slope of a vertical line is undefined. Annette’s indifference curves endlessly rise and have no run, so the negative of an undefined number (the slope) remains an undefined number (the mrs). Her indifference map therefore represents a range of vertical lines where the horizontal intercepts correspond to the amount of money she keeps which is also the utility number associated with the particular indifference curve. 151 152 MICROECONOMICS - DRAFT Now, we might ask ourselves, what is Annette’s utility at her constrained utility maximum? Let’s substitute in the values we have for p A = 10 and p B = 0. uA (p A , p B ) = (p A )1 (p B )0 = (10)1 (0)0 = 10 (3.21) Annette has a utility that is equal to the amount of money she keeps for herself. M-Note 3.11: An altruistic person splitting the pie We will derive Chen’s decision about splitting the pie between him and Diane. Using Equation 3.10, his marginal rate of substitution is (see his utility, Equation 3.19): mrs(p C , p D ) 7p D 3p C = Now, let’s assume that the size of the pie is z = 1, therefore, the feasible allocations are represented by p D = 1 p C , so his mrt (the negative of the slope of the feasible frontier) is mrt = dp D dp C = 1 Equating the mrs with the mrt , we can obtain how much Chen allocates to himself and to Diane: 7p D 3p C = 1 pD = 3 C p 7 = 1 = 1 = 0.7 = 0.3 3 C p 7 10 C p 7 ) pC Using the feasible allocation set and pD pC That is why Chen offers Diane $3 of the total of $10 that she is able to allocate. Checkpoint 3.9: Chen’s Choice and the mrs = mrt rule. 1. What is the marginal rate of transformation in this in the game described in Figure Figure 3.14? 2. Why is the utility maximizing outcome bundle at point b in Panel b of Figure 3.14 an example of a case there the mrt = mrs rule does not work. How does this case differ from the case shown in Panel b, where the rule does work? 3. Use the value of Chen’s mrs at point c’ in Figure 3.14 Panel b along with the value of the mrt to explain why for Chen the opportunity cost of giving more money to Diane is less than his willingness to pay (give up his own payoffs) so that Diane can have more. 3.11 Application: Environmental trade-offs We think of environmental damage as something to be avoided, but stopping or slowing the damage – or "abating" the damage in the language of environ- D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 153 mental science – is costly. Less damage means some combination of less consumption, changing our consumption patterns to be less damaging to the environment, or diverting our productive potential from producing goods that we can now consume to discovering and installing new technologies. We therefore face a trade-off between consuming goods and maintaining the quality of the environment. How much of these opportunity costs of improved environmental quality are we willing to pay? The constrained utility maximization method we have developed provides a way of posing and answering these questions using the preferences, beliefs, and constraints approach. Feasible combinations of conventional goods and environmental quality The opportunity cost of environmental quality is consumption of other (conventional) goods such as food, clothing, shelter, and transportation, which we must give up to secure a higher quality environment. There is a feasible frontier showing the combinations of environmental quality, x, and conventional goods, y, that are possible for a society. The feasible frontier in the case of environmental quality depends on the abatement technology, which represents how much consumption of conventional goods society has to give up to achieve a given level of environmental quality. Figure 3.15 shows a feasible frontier between conventional goods (y) and environmental quality (x). We measure environmental quality on a numeric scale from 0 (the environment that we would have if no abatement done) to 20 (the environment resulting if we were to divert to abatement uses all of society’s resources above some minimum level of consumption). We measure conventional consumption as billions of dollars. The negative slope of the feasible frontier at any point is the marginal rate of transformation of reduced environmental quality into increased conventional consumption, or Dy Dx . The steeper the frontier, the greater is the increase in feasible consumption allowed by a reduction in environmental quality. This is also the the opportunity cost of improved environmental quality. So a flatter frontier means a lower opportunity cost of abatement. To see this, starting at no abatement expenditures (y = ȳ), the opportunity cost of improved environment is initially small (the frontier is nearly flat) and as Annette implements more abatement, the cost more abatement increases as the environmental quality increases. The shape of the feasible frontier reflects an increasing marginal rate of transformation, or an increasing marginal opportunity cost of environmental quality. Put another way, if environmental quality is at its maximum at the intercept of the feasible frontier with the horizontal axis, society could consume a lot H I S TO RY In the middle of the twentieth century, long before we worried about climate change and its unfolding calamities, Aldo Leopold, the American environmentalist raised an economic question: "Like winds and sunsets, wild things were taken for granted until progress began to do away with them. Now we face the question whether a still higher ’standard of living’ is worth its cost in things natural, wild and free." (Leopold 2020 [1949], p. xxi). Leopold was articulating a trade-off between, on the one hand, consuming goods and services – Leopold’s higher "standard of living" – and on the other, the costs of environmental damage – the "cost in things natural, wild and free." 154 MICROECONOMICS - DRAFT Figure 3.15: Trade-off between consumption of conventional goods and environmental quality. The constrained utility maximum is the point on the feasible frontier on the highest indifference curve u2 , shown as point b where the mrs = mrt rule holds. The constrained maximum is at the point where the feasible frontier is tangent to the highest attainable indifference curve. Policy−maker's indifference curves Goods in millions, y y Abatement Cost yb b u3 u2 u1 Initial feasible frontier xb x Environmental quality, x more conventional goods if it were willing to tolerate a small deterioration of environmental quality (the frontier is steep where it intercepts the horizontal axis). But the feasible increase in consumption of conventional goods allowed by a reduction in environmental quality falls as the level of environmental quality declines. Checkpoint 3.10: The mrt of the environmental feasible frontier a. Refer to Equation 3.22 and find the marginal rate of transformation of the feasible frontier. b. Practice sketching the feasible frontier and confirm the intercepts with the horizontal and vertical axes as shown in Figure 3.15. 3.12 Application: Optimal abatement of environmental damages How much abatement is the right level, taking account of both preferences for conventional goods (consumption) and the quality of the environment along with the opportunity costs in lost consumption? A citizen chooses a level of abatement of environmental damages To begin with the simplest case, think of just one citizen, Annette, who might be representative of the attitudes of the whole society, trying to decide on the M - C H E C K An example of the function representing the feasible frontier is shown below, where y is the goods available for consumption, ȳ is the level of y that is feasible when environmental quality is at its minimum, and x is environmental quality. y = 100 1 2 x 2 (3.22) This is the equation for the feasible frontier is graphed in Figure 3.15. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N level of abatement that she would like to see implemented. She cares about both the quality of the environment, x, and the amount of conventional goods that will be available for people to consume Annette’s utility function has the following form: u = u(x, y). Annette considers what she would like to see her society do about the environment (x), taking account of the effects on everyone. In other words, she is thinking from an other regarding like an ideal policy-maker. Annette’s indifference curves between environmental quality and conventional goods are downward sloping because she regards both environmental quality and conventional consumption as goods for which more is better. This means the marginal utility of both y and x are positive (e.g. uy > 0 and ux > 0). The negative slope of the indifference curves shown in Figure 3.15 at any point is Annette’s marginal rate of substitution between more consumption of goods and a better environment. Her marginal rate of substitution shows the amount of goods she would be willing to give up for a small improvement in the environment. As before, Annette’s indifference curves exhibit diminishing marginal utility of both environmental quality and consumption. An example of a utility function that Annette might have is the Cobb-Douglas utility function: u(x, y) = xa y(1 a) = x0.4 y0.6 (3.23) Figure 3.15 shows three indifference curves defined by equation 3.23: uA 3 is unattainable given the feasible frontier, uA 1 intersects the feasible frontier twice, and uA 2 is tangent to the frontier at point (xb , yb ). Annette’s constrained maximum allows her and her fellow citizens to consume 75 million units of conventional goods and enjoy environmental quality of about 7 (see M-Note 3.12 for the worked solution). If she were able to implement relevant environmental and fiscal policies, this point is the best society can do in Annette’s opinion. What is the total opportunity cost in foregone conventional consumption of a level of environmental quality of 7? The maximum feasible level of conventional consumption with no abatement is $100 billion. The difference between the maximum feasible consumption of $100 billion and Annette’s preferred choice of conventional consumption of $75 billion is the opportunity cost of an environmental quality of 7. In our example, the abatement costs are equal to $100 billion $75 billion = $25 billion in conventional goods. A citizen with Annette’s preferences thinks that the sacrifice of $25 billion consumption goods is more than worth paying to have an environmental quality of 7 instead of zero. 155 156 MICROECONOMICS - DRAFT New technologies, and conflicts of interest If, with a mind to the future, some of the abatement costs are devoted to research to improve abatement technologies, this would pivot the feasible frontier outwards, as shown in Figure 3.16. Remember this is very similar to the expansion of the feasible set shown in Figure Figure 8.3 when Keiko adopted improved studying so as to reduce the opportunity cost in reduced Living time associated with greater Learning. As is shown in the figure, the shift of the feasible frontier would permit higher environmental quality of x ⇡ 9.8 at the same level of consumption of $75 billion at the new (xr , yr = yb ). But there would still be a trade-off: more conventional goods would require less environmental quality, or more environmental quality would require fewer conventional goods to stay on the feasible frontier. We can also use Figure 3.15 to see why people often disagree about environmental policy. • Preferences: peoples’ preferences for conventional goods and the environment may differ • Beliefs: people may disagree about the opportunity costs or the benefits of environmental quality • Conflicts of interest: the costs and benefits of abatement fall on different people; those whose jobs or profits depend on carbon based energy, for example, stand to bear more of the costs of addressing climate change, while regions likely to be particularly hard-hit like Africa bear a larger share of the benefits. M-Note 3.12: The trade-offs and opportunity costs of the environment Let us work through the process that Annette the policy-maker would go through to identify the combination of goods in billions of dollars with environmental quality. First, let us calculate her marginal rate of substitution from her utility function, uA (x, y) = (xA )0.4 (yA )0.6 . From earlier in the chapter, we know that the mrs(x, y) is the ratio of marginal utilities and we have already calculated this for a = 0.4 and (1 a ) = 0.6 in Equation 3.13 in M-Note 3.7. mrs(x, y) 2y 3x = (3.24) Annette’s feasible frontier, based on her beliefs and understanding of the existing science, dy is given by the equation y = 100 12 x2 , for which we can find her mrt (x, y) = dx : ) dy dx dy dx x = = x (3.25) We now set the mrt (x, y) given by Equation 3.25 equal to the mrs(x, y) given by Equation F AC T C H E C K The pace of environment friendly innovation is astounding. Have a look at the reduction in costs of the photovoltaic cells used in solar panels dropping to one-onehundreth of there costs in 1975 in Figure 8.3. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 3.24 and we isolate one of the variables, y: Multiply through by 3x Divide through by 2 x = 3x2 = y = 2y 3x 2y 3 2 x 2 (3.26) We can now substitute Equation 3.26 into the feasible frontier to find xb and yb : 3 2 x 2 2x2 = 100 1 2 x 2 = 100 x2 = ) xb = 50 p 50 = 7.07 Having found xb , we can substitute it back into 3.26 to find yb : y y b = 100 = 100 = 75 1 p 2 ( 50) 2 1 (50) 2 So, as a result of Annette’s policy-making utility function and feasible frontier, she would choose a combination of environmental quality, x, of value 7.07 with consumption of good and services of $75 billion. $75 billion is $25 billion less than the maximum consumption of goods and services, ȳ = 100, so the cost of abatement is $25 billion. First, people may differ in their preferences over conventional goods and the environment. Another citizen, Brenda, may not worry as much as Annette about the climate and problems that future generations will inherit because she puts a lower weight on the welfare of others (in this case of future generations). Brenda would have different indifference curves. If she also had Cobb-Douglas utility, she would have less strong preferences for the environment than Annette, with a lower a and higher (1 a ). Brenda would as a result choose a constrained maximum with lower social spending on abatement. A second reason for disagreement is that Brenda thinks that the costs of abatement are much greater than Annette thinks they are. If Brenda thinks that the actual costs of environmental quality are greater than Annette does, Brenda would work with a different feasible frontier inside the feasible frontier Annette considers. Brenda’s feasible frontier would be steeper over its entire range, indicating that the opportunity cost of environmental quality is higher for any level of environmental quality than on Annette’s feasible frontier. Brenda’s constrained utility-maximizing bundle would be different from Annette’s. But there is a third reason for disagreement, not having to do with the preferences or beliefs of the citizens. Some members of the society may benefit from decisions that harm the environment. Brenda might be employed by or 157 158 MICROECONOMICS - DRAFT Figure 3.16: Trade-off between consumption of conventional goods and environmental quality with R&D. The choice between consumption of conventional goods and environmental quality with R&D pivoting the feasible frontier outwards leading to a new point of tangency on a higher indifference curve u3 . Policy−maker's indifference curves Goods in millions, y y Abatement Cost yb r b u3 u2 u1 Initial feasible frontier xb xr Feasible frontier with R&D x Environmental quality, x an owner of a firm producing fossil fuels, for example. They will bear more than proportionally the costs of abatement and may prefer lesser levels of abatement for that self-regarding reason. The fact that a policy of CO2 emissions abatement would affect Brenda adversely while benefiting Annette brings us back to the how we understand the term utility. Checkpoint 3.11 a. Show that when Annette’s utility function is defined by Equation 3.23 and the feasible frontier is defined by Equation 3.22 with ȳ = 100, that the utilitymaximizing consumption bundle is $75 billion with an environmental quality of 7.07. b. Show that when Annette’s utility function is defined by Equation 3.23 and the feasible frontier is defined by Equation 3.22 with a technological im1 2 provement where y = 100 3 x that the constrained utility-maximizing consumption bundle is (xr = 8.66, yr = $75 billion). c. Show that when Annette’s utility function is defined by Equation 3.23 and the feasible frontier is defined by equation 3.22 with a technological improve1 2 ment where y = 100 4 x , the constrained utility-maximizing consumption bundle is (xr = 10, yr = $75 billion). Draw a new feasible frontier tangent to a new indifference curve u4 with accurate intercepts reflecting the new technology. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 3.13 159 Cardinal inter-personally comparable utility: Evaluating policies to reduce inequality Most policy choices involve conflicts of interest like that between Annette and Brenda about the abatement of environmental harms. Few policy choices are entirely win-win. Most policies – whether they concern taxation, immigration, health insurance, or the rate of inflation – result in benefits for some and losses for others. Ordinal and cardinal utility in policy evaluation How do we then evaluate competing policies? Don’t think about this as a question about what would be a good outcome for you if you were a participant in the society. Instead, try to take the position of what Adam Smith called the Impartial Spectator who did not himself stand to gain or lose, but wanted instead to consider the gains and losses to society. One answer you might give is just to count those who prefer each policy and select the most popular policy. All this requires is that people be able to rank H I S TO RY Adam Smith in The Theory of Moral Sentiments conceived of the impartial spectator as follows, "We endeavour to examine our own conduct as we imagine any other fair and impartial spectator would examine it. If, upon placing ourselves in his situation, we thoroughly enter into all the passions and motives which influenced it, we approve of it, by sympathy with the approbation of this supposed equitable judge. If otherwise, we enter into his disapprobation, and condemn it."3 the policies in question as better, worse, or indifferent. We could in this case treat utility as ordinal (that is, simply a ranking (or ordering) of outcomes). Something like this might occur in a majority rule democratic political system, especially if citizens could vote on policies as they do in many countries in referendums asking citizens to vote for or against a particular policy. But this way of evaluating policies might result in evaluating positively those policies that confer minor gains to those in favor, and substantial losses to those preferring another policy. This does not seem like a sensible rule. An alternative is to weigh the amount of the gains to the beneficiaries of each policy against the size of the costs incurred by those who would have done better under some other policy. This kind of comparison requires that we know not only which policies people prefer, but how much they prefer them. To do this we treat utility as a cardinal measure for which utility is not just a ordinal ranking, but instead a number indicating how well off the person is under the option in question. Treating utility as cardinal allows us to say two very different things: 1. for Annette, the outcome (x0 , y0 ) is twice as good as (x, y) because for example uA (x0 , y0 ) = 2uA (x, y) 2. the sum of the Annette’s and Brenda’s utility is greater with outcome (x0 , y0 ) than with outcome (x, y) because uA (x0 , y0 ) + uB (x0 , y0 ) > uA (x, y) + uB (x, y) Both statements involve cardinal utilities, but they differ. The first statement compares how much Annette values two different states that she will experi- H I S TO RY Lionel Robbins (1898-1984) was a leader in the "ordinal revolution" in economics. Economics, he wrote, does not need "to compare the satisfaction which I get from the spending of 6 pence on bread with the satisfaction which the Baker gets by receiving it. That comparison . . . is never needed in the theory . . . ." (123). Moreover, "There is no way of comparing the satisfactions of different people" (124). 160 MICROECONOMICS - DRAFT ence. It does not compare her evaluation of a state that she will experience with someone else’s evaluation of the state they will experience. The first statement is an example of the cardinal utility that we introduced in Chapter 2 as the basis of expected payoffs (or expected utility) and the analysis of decision-making in risky situations. The second statement compares Annette’s utility with Brenda’s utility. When utility is represented in this way it is called inter-personally comparable cardinal utility (or sometimes “cardinal full comparable utility”). If utility is cardinal in this inter-personally comparable sense, then we can compare how well off two or more people are, and how much better off or worse off a policy would make each of them. This provides a way to evaluate which policies should be implemented by asking whether H I S TO RY Philosopher-economist John Stuart Mill (1806-1873) referred to what we would now call the sum of the total utilities of a population as as "a good" that should be promoted: "the general happiness is desirable... each person’s happiness is a good to that person, and the general happiness, therefore, a good to the aggregate of all persons."4 the gains of those who benefit from a policy exceed the losses of those who do not. Why do these two methods of comparing utility matter? Remember that one of the problems with Pareto efficiency as a criterion for fair policy outcomes was 6 an adequate basis for an Impartial Spectator preferring one outcome over the 5 other. Using the second – stronger – conception of cardinal utility along with 4 the judgement that one outcome is better than another if total utility is greater provides a rule for evaluating which Pareto-efficient outcome we might prefer as a society. Bob's payoffs many outcomes can be Pareto efficient. So Pareto efficiency does not provide d 3 c 2 a 1 In the payoffs for the Fishermen’s Dilemma in Figure 1.11 (shown here in the margin for easy reference) three of the four outcomes of the game are Pareto efficient. The Pareto criterion provides no way to choose among them. By contrast the rule — maximize total utility – selects the mutual cooperate b 0 0 1 2 3 4 5 6 Alfredo's payoffs Figure 3.17: Pareto comparisons in the Fisherman’s Dilemma Game. The Pareto criterion favors c over a (which it dominates) but cannot rank points b, c and d because are all Pareto efficient. outcome (point c) with total utility of 6. Cardinal utility and the distribution of wealth Suppose you are a policy-maker and you have to divide an amount of wealth between Annette and Brenda. The amount of wealth you have to divide is equal to 1, so each person can get a fraction of that wealth and, as long as the fractions sum to 1, then the outcome will be Pareto-efficient. Let the fraction that Annette gets be a with Brenda’s fraction be (1 a). Annette and Brenda have identical preferences for wealth given by the cardinal utility functions of how much wealth they get: uA (a) and uB (1 a). For both of them the marginal utility of wealth is diminishing with increased wealth. The horizontal axis in Figure 3.18 shows all possible distributions of wealth between Annette and Brenda. • Annette’s share of wealth, a, varies from 0 to 1. • At a = 0, Annette gets nothing and Brenda gets everything. E X A M P L E When you say "I’ll do the shopping; it’ll be less trouble for me than for you" you are representing utility (the trouble of shopping) as cardinal and making an interpersonal comparison of utility (less trouble for me than for you). In fact, much of our everyday ethical reasoning involves interpersonal comparisons of the benefits and costs that people experience. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N A's marginal utility uAa Marginal utility muB(1 − ah) Figure 3.18: Distribution of wealth, marginal utility, and total utility. In the figure a is the proportion of wealth belonging to Annette (A). 1 a is the proportion of wealth belonging to Brenda (B). As a person’s wealth increases, the marginal utility of wealth decreases. A’s total wealth increases as you move along the bottom line from left to right, and as a result her marginal utility decreases. Because B’s wealth increases as the division moves towards the left, B’s marginal utility decreases from right to left. B's marginal utility − uBa g i h muA(ah) 0 ai 0 ah 1 A's share of wealth, a • At a = 1, Brenda gets nothing and Annette gets everything. Figure 3.18 shows the two marginal utility functions for wealth. Each of them has decreasing marginal utility in wealth. This means that the increment in utility associated with each additional unit of wealth they have is less when they have more wealth. Annette’s marginal utility curve slopes downward as she gets more wealth (moving from left to right), and Brenda’s marginal utility of wealth decreases as she gets more wealth (moving from right to left). Suppose the status quo is ah , a situation in which Annette is wealthy and Brenda is poor (Annette’s has share of wealth ah and Brenda’s share of wealth is 1 ah ). A policy that takes a small amount of wealth from Annette and transfers it to Brenda reduces Annette’s utility by less than it increases Brenda’s. We can see this by identifying that Annette’s marginal utility at ah , uAa (ah ) is much lower than Brenda’s utility at the same point uBa (ah ). The vertical difference between points g and h shows the magnitude of the difference in their marginal utilities. Redistributing wealth from Annette to Brenda therefore increases total utility (the sum of Annette and Brenda’s utilities). Applying this reasoning to other points in the diagram, we find that the distribution of wealth that maximizes total utility is ai , where Annette’s marginal utility of wealth equals Brenda’s marginal utility of wealth. If Annette’s and Brenda’s utility functions are identical, the total utility maximizing point dis- 161 162 MICROECONOMICS - DRAFT tributes wealth equally, ai = 12 . M-Note 3.13: Maximizing Total Utility Consider a society of two people, Annette (A) and Brenda (B), in which maximizing total utility, U = uA + uB , is the goal of the policy-maker who we will assume is Adam Smith’s Impartial Spectator . The Impartial Spectator selects a point, i, for their choice of policy. We assume Annette and Brenda assign the same utility numbers to some given level of wealth. They are identical in this respect. But they differ in their wealth. Annette’s share of total wealth is a > 12 and Brenda’s is 1 a < 12 < a. Annette’s utility is uA = u(a) and Brenda’s utility, uB = u(1 a), is also a function of a, but Brenda’s utility decreases as a increases. U (a) uA + uB = u(a) + u(1 = a) (3.27) To find the maximum total utility, we differentiate the total utility with respect to a and set the derivative equal to zero: dU da ) ua (ai ) Ua (a) = = uAa + uBa = ua (a) = ua (1 = a) = 0 ai ) The only way that ua (ai ) can be equal to ua (1 ber. ai ua (1 1 ai ) is if ai and 1 ai = ai are the same num- 1 2 This maximization is depicted in Figure 3.18: Annette’s marginal utility, uA a (a), decreases as a increases (diminishing marginal utility) and Brenda’s marginal utility uB a (1 creases as a increases (because her wealth, 1 a) in- a decreases as a increases). The total utility maximizing choice occurs at ai , where wealth is equal and as a result the marginal utilities are equal: uA (ai ) = uB (1 3.14 ai ). Application: Cardinal utility and subjective well-being A century ago economists thought that while ordinal comparisons like better or worse are possible empirical interpersonal comparisons expressed by a number indicating the degree of preferences were impossible to make. But today researchers are actively engaged in measuring individual happiness and life satisfaction, using techniques ranging from surveys and natural observation to E X A M P L E Daniel Kahneman, a psychologist and Nobel Laureate in economic, has advocated a hedonistic (meaning concerning pleasure and pain) theory of utility. Kahneman titled one of his papers “Back to Bentham?" to pay homage to the early 19th century philosopher economist Jeremy Bentham’s utilitarian theory.5 the methods of experimental neuroscience. They are asking such questions as: "How important is income for happiness?" "Is being without a job a bigger source of unhappiness than being without a spouse?" These researchers refer to happiness or life satisfaction as subjective well-being. To measure "pleasures and pains" in the lab, volunteers are exposed to an electrical shock and asked to report on their experience of that on a numerical scale. Others are asked to plunge their hands into extremely cold water for as long as they can stand it and immediately report their level of unhappiness having done so. Respondents in surveys are asked their "life satisfaction." This research has sought to understand the activities that make people most F AC T C H E C K The Satisfaction with Life survey is based on five questions each of which is rated on a 7-point scale from Strongly Disagree (1) to Strongly Agree (7). Here are the questions: In most ways my life is close to my ideal; The conditions of my life are excellent; I am satisfied with my life; So far I have gotten the important things I want in life; and If I could live my life over, I would change almost nothing. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 163 happy. Almost all people surveyed seem to like sex quite a lot, ranking "intimate relations" as having a high subjective well-being value. Ranked after sex, people like socializing, relaxing, sharing meals with friends, praying, and exercising. People don’t like housework, childcare, commuting or working. People also report major changes in subjective well-being from painful events, like sudden loss of a job, a death in the family, or divorce, or from positive events like marriage, or the birth of a child.6 But when you ask someone about their happiness over time the measures are surprisingly consistent: people are likely to report similar activities or outcomes as providing them with happiness when you ask them at different intervals. What are the take-home messages about subjective well-being, the choices about how we spend our time, and how we value effortful work? First, people like a diverse array of activities and doing things they like provides them with happiness that can be measured in meaningful ways across different people and at different periods of time in our lives. F AC T C H E C K Non-laboratory measures of subjective well-being suggest that people with higher subjective well-being tend to be less likely to contract a cold virus and to recover more quickly when they do contract the cold. Similar evidence exists for people who have recovered from wounds and had baseline and subsequent subjective wellbeing measured: those who are happier recover more quickly.7 Second, people who report greater subjective well-being are also better off by physical biological measures. For example,they are less likely to be ill. Subjective well-being also manifests in hormone levels, brain patterns, and palm temperature.8 Third, while income matters for happiness (especially for people without much income) people value social relationships – marriage, a job, friendships – more than they value income.9 Making the transition from unemployed to employed boosts a persons subjective well-being by much more than would be predicted simply by the increase in income. This is because having a job is a source of respect and dignity, especially as it provides a way for people to express autonomy over themselves, competence in their expression of their abilities, and relatedness to other co-workers and people around their work. Checkpoint 3.12: Joy or Misery? Think about the kinds of activities that Kahneman and Krueger discuss above that provided people with joy (that they ranked highest in terms of providing them with subjective well-being). 1. Compare them with their opposites: those that result in disutility or even misery. 2. Come up with a list of activities that you engage in that provide you with joy which you try to prioritize. 3. Why do you spend the time that you do on these activities? Why do you not spend more? 4. Do you engage in activities that in the moment are unpleasurable but which you believe provide you with benefit nonetheless? E X A M P L E The substantial subjective cost that people experience when they are out of work is one reason why employers (who have the power to terminate a person’s job) have power over their employees. We shall return to this when we study the firm and the labor market. 164 MICROECONOMICS - DRAFT 5. Do you think such activities appear in the models we’ve developed? 3.15 Preferences, beliefs, and constraints: An assessment Many scholarly disciplines in addition to economics are devoted to understanding human behavior including psychology, sociology, anthropology and history, but also more distant endeavors including literature, philosophy, neuroscience, computer science and biology. The preferences, beliefs and constraints approach, while a standard set of tools in economics that is widely used in other fields, is just one of many approaches. People newly familiar with the approach often raise the following questions about it. • Are people really all that selfish? This concern is based on a misunderstanding of the model,which says nothing about whether people are seeking to help others, aggrandize themselves, or a little of both. Our treatment of altruism, reciprocity and fair-mindedness shows that the model – using indifference curves and feasible sets, for example – can apply to a variety of motives. • Do people consciously optimise, for example, applying the mrs = mrt rule when they shop? The model is not a description of how people actually think or their emotional states when they take a break from studying, or support a particular environmental policy. We model instead what people would do if they did the best that they could. The fact that the model often yields predictions similar to what we observe empirically (including by experiments, econometric and other quantitative methods) does not require that the model is an accurate representation of the process by which people come to take one course of action over another. In some cases, people consciously optimize, going though mental calculations similar to the model. For example, a person buying a house or choosing between two job offers will weigh the pros and cons of the alternatives. But in other cases, the actions may not even appear to us as a decision, for example, what to eat for breakfast, what to wear today, or what our personal values should be. Without consciously trying to do so, people may arrive at something like the solution to these optimization problems by trial and error, or by observing others who seem to be successful or happy with their choices, or by following habits that will remain in place unless changed by some dramatically adverse consequences of following them. Other concerns about the model are more serious. • What about emotions and visceral reactions, aren’t they important? This question points to a shortcoming of the approach; but it is not that the approach excludes emotions like fear, shame, and attraction. The shortcoming is that the preferences, beliefs, and constraints approach says almost H I S TO RY In his 1953 work,Essays in Positive Economics, Nobel Laureate Milton Friedman (1912-2006) observed that "predicting the shots made by an expert billiard player" could be done on the basis of "the complicated mathematical formulas that would give the optimum directions" of the shots. But this prediction would not be "based on the belief that billiard players, even expert ones, can or do" actually make these calculations.10 D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N 165 nothing at all about motives, that is, it says nothing about the reasons why people rank some outcome as superior to another. Knowing more about motives like this would help us understand economic and other behavior. • Preferences and beliefs are not "just there" as facts of nature, they are products of environments we live in. We have already seen an example of this as social norms concerning the kinds of work that are appropriate for women to do changed during the 20th century under the influence of new technologies for cooking cleaning and washing and the experience of women doing factory work producing armaments during the second world war. Our preferences and beliefs are to some degree "socially constructed". Our discussion of differing cultures around the world using experiments designed using the preferences, beliefs and constraints framework shows that the approach can help to clarify how cultures differ and how society shapes preferences and beliefs. • Commitments and consequences. The framework is based on the idea that our behavior is based on our beliefs about the consequences our actions will bring about in the future. Don’t we sometimes act to fulfill promises or other commitments made in the past, or just to "do the right thing" without regard to future consequences? Yes we do, and a shortcoming of the approach is that it does not address that kind of behavior. • Predicting behavior and evaluating outcomes. Economists use the same concept "utility" in models designed to predict the actions that people will take and to provide the basis for evaluating economic outcomes and public policies to improve them. The idea is that whatever it is that motivates people to make the choices they do should also be the objective of public policy and form the basis for our preferring one societal outcome over another. But treating actual behavior as if it were the pursuit of a concept of well-being that should be the basis of our judgement of societal outcomes is a mistake. The reasons for our actions (that is, our preferences) include addictions, weakness of will, shortsightedness, and other well documented socially dysfunctional aspects of human behavior that in retrospect are often deeply regretted by those acting on them. A sensible conclusion from reviewing these concerns about the preferences beliefs and constraints approach might be that the approach is better for answering some questions than others, and learning to distinguish which is which is an important learning objective. As we said at the beginning of the chapter: the map is not the territory. Good maps don’t have all the information about the territory they depict and good economic models require us to leave some things out. Checkpoint 3.13: Positive and normative uses of "utility" H I S TO RY : P O S I T I V E A N D N O R M AT I V E E C O N O M I C S The distinction between the economics of "what is" called positive economics and "what ought to be" called normative economics was made by John Maynard Keynes in his 1893 Scope and Method of Political Economy and by Milton Friedman in his 1953 The Methodology of Positive Economics. The distinction is controversial in part due to differences about the appropriate role for "what ought to be" statements in economics.11 166 MICROECONOMICS - DRAFT Consider the statement by J.S.Mill (in the above margin note): "each person’s happiness is a good to that person, and the general happiness, therefore, a good to the aggregate of all persons." Explain how Mill is here using "happiness" both as a way of predicting behavior (sometimes called "positive economics") and as a way of evaluating outcomes from a societal standpoint ("normative economics.") 3.16 Conclusion In this chapter we have studied the constrained optimization problems shown in Table 3.1. Though the problems concerned are quite different, the models and analytical tools we used to analyze them are very similar. In each case the analysis of the decision involves two kinds of trade-offs: • The first trade-off that appeared in each of these situations is the actor’s relative valuation of the things she cares about, measured by the negative of the slope of an indifference curve, that is, the marginal rate of substitution. • The second trade-off is that at any point on the feasible frontier, the opportunity cost of having more of one good that the actor values is that she must have less of another good that she values. This opportunity cost trade-off is measured by the negative of the slope of the feasible frontier, that is, the marginal rate of transformation. The result - the action taken doing the best she can under the constraints she faces - is determined in the same way in all the cases: by finding the point on the feasible frontier that is on the highest indifference curve. This will often be the bundle where the mrs = mrt rule holds. The table demonstrates that many seemingly different kinds of action can be studied with a common model, one that we will use often. In this chapter, we have focused on single actors and for the most part excluded from the model something important: other people. With the exception of the farmers of Palanpur, we have modeled the person facing a given situation defined by a feasible frontier and preferences represented by indifference curves. (We already explained that the second person in the Dictator Game is not really a player at all). We now turn to a world populated by people interacting strategically, and we ask how economic institutions affect the outcomes of these interactions, and how these outcomes and institutions might be judged by the standards of Pareto-efficiency and fairness. We shall continue to employ the tools of constrained utility maximization: we will continue to understand people’s tradeoffs through their marginal rates of substitution and of transformation; and we shall need these to understand how one person might engage in exchange of D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N Actor Utility depends on Action Constraints Keiko Learning, Living time Time allocation Learning-Living feasible frontier Keiko Learning, Study effort Study effort Study-Learning production function Annette/Chen Payoffs to two players Payoffs to each The total endowment Annette/Brenda Conventional goods, Conventional goods Consuming conventional goods Environmental quality 167 degrades the environment goods and services with someone else. Making connections Strategic and non-strategic social interactions: In the previous chapters we considered strategic social interactions – like the fishers and the farmers from Palanpur. Here we look at simpler aspects of behavior when a person is attempting to do the best they can in situations that are not strategic because the choice of how hard to study, or how much fish to buy is not greatly affected by others choices. Self-regarding and social preferences: In Chapter 2 we provided evidence that people can be self-regarding, altruistic, reciprocal, and fair-minded. These diverse behaviors can be modeled using the preferences, beliefs, and constraints approach using indifference curves and feasible frontiers, as we showed for the case of an altruist. Opportunity costs and trade-offs: Regardless of whether a person’s preferences are self-interested or social, people face trade-offs among the ends they wish to pursue and they face opportunity costs when trying to choose a course of action. Public policy: Economics engaged: The idea of constrained utility maximization illustrated the trade-off between consuming more goods on the one hand or either consuming less and using some of the economy’s resources to abate environmental damages, obtaining greater environmental quality. We also modeled the choices an altruistic person might make in sharing something of value thereby providing a model capable of analysing the kinds of result observed in the experiments reviewed in the previous chapter. Evaluating outcomes: Treating utility as cardinal and inter-personally comparable rather than ordinal allows us to compare the benefits and burdens that a policy will impose on different people. This provides a basis (one of a number of alternatives) for saying that one policy or outcome might be preferred to another, as illustrated by the case of the distribution of wealth. Table 3.1: The constrained optimization problems used in this chapter 168 - DRAFT MICROECONOMICS Important ideas preference constraints beliefs Homo economicus altruist reciprocator ordinal utility cardinal utility utility function and value function total utility marginal utility Cobb-Douglas utility mrs law of diminishing marginal utility mrt rule slope indifference curve marginal rate of substitution diminishing marginal rate of substitution tradeoff willingness to pay iso-value curve feasible/attainable feasible frontier production function marginal product marginal rate of transformation increasing marginal rate of transformation opportunity cost increasing opportunity costs utility-maximizing point of tangency price line offer curve Mathematical notation Notation Definition u() x y h a ȳ c() a p u z p utility function a good (or a "bad") a good (or a "bad") hours of work Cobb-Douglas exponent of good x vertical intercept of the feasible frontier opportunity cost A’s share of wealth price of a good constant utility along an indifference curve endowment in the Dictator game payoff in the Dictator game Note on super- and subscripts: A, B, C, D: different people; CD: CobbDouglas utility function; Subscript b indicates where someone does the best they can; RD: feasible frontier with Research and Development. Discussion questions See supplementary materials. Problems See supplementary materials. D O I N G T H E B E S T YO U C A N : C O N S T R A I N E D O P T I M I Z AT I O N Selected Answers/Hints for Questions See supplementary materials. 169 4 Property, Power & Exchange: Mutual Gains & Conflicts DOING ECONOMICS [T]he efforts of men are utilized in two different ways: they are directed to the production or transformation of economic goods, or else to the appropriation of goods produced by others. Vilfredo Pareto, Manual of Political Economy (1905) (Pareto (1971):341) Ibn Battuta, the fourteenth century Moroccan scholar, reported that along the Volga River in what is now Russia, long distance trade took the following form: “Each traveler ... leaves the goods he has brought ... and they retire to their camping ground. Next day they go back to ... their goods and find opposite them skins of sable, miniver, and ermine. If the merchant is satisfied with the exchange he takes them, but if not he leaves them. The inhabitants then add more skins, but sometimes they take away their goods and leave the merchant’s. This is their method of commerce. Those who go there do not know whom they are trading with or whether they be jinn [spirits] or men, for they never see anyone." The Greek historian Herodotus describes similar exchanges between Carthaginian and Libyan groups in the 5th century B.C. After having left their goods, Herodotus reports, the Carthaginians withdraw and the Libyans “put some This chapter will enable you to do the following: • Explain why, when people exchange goods, there are both mutual gains and also conflict over the distribution of these gains. • Understand how an allocation of goods can be evaluated on grounds of Paretoefficiency and fairness. • Show how self-regarding as well as social preferences of parties to an exchange can affect the outcome, and how other-regarding social preferences may reduce the scope of conflicts over the distribution of the gains from exchange. • Understand how both private property rights and the exercise of power by one of the parties to an exchange will affect the outcome of exchange. • Use the ideas you have learned to explain how an employer and an employee might bargain over working hours and wages. gold on the ground for the goods, and then pull back away from the goods. At that point the Carthaginians ... have a look, and if they think there is enough gold to pay for the cargo they take it and leave.” Herodotus describes how the process continues until an acceptable is price hit upon, remarking with surprise that “neither side cheats the other ... [the Carthaginians] do not touch the gold until it is equal in value to the cargo, and [the Libyans] do not touch the goods until the Carthaginians have taken the gold.” Alvise da Ca da Mosto, a fifteen century Venetian working for the Portuguese crown, reported a similar practice in the African kingdom of Mali, regarding it as “an ancient custom which seems strange and hard to believe.” Figure 4.1: A painting of Ibn Battuta (on the right) (1304-1369) Source: Wikimedia Commons. 172 MICROECONOMICS - DRAFT But is the so-called ‘silent trade’ really so odd? Transfers of goods among strangers can be dangerous. What one expected to be an exchange at mutually agreeable prices may end up as theft or an “offer you cannot refuse.” But trade among strangers can also be highly profitable. The potential gains from trade are often greater the more distant geographically or socially the parties are to the exchange: the salt brought by Tuaregs from the Atlas Mountains in North Africa across the Sahara by camel to the Kingdom of Ghana was not available in West Africa. The gold and tropical nuts Tuaregs gained in silent trade with Ghanaians was not available north of the Sahara. The silent trade – with its unusual etiquette in which parties interacted only at a distance – allowed both Tuaregs and Ghanaians to get some of what they lacked and wanted in return for giving up some of what they had in abundance and could readily part with. They were exploiting the mutual gains that differences in geography, tastes, technologies, and skills allow. And the rules of the game for governing their exchange process – the institutions that we call "the silent trade" were a way of doing this and dividing the mutual gains without resorting to violent conflicts. Other than these mutually advantageous exchanges, there are many other ways that goods change hands: from the use of violent coercion by private parties (i.e. theft), or by the use of one nation’s military force to acquire the resources of another people. People have also been violently coerced into work through enslavement by private actors and states alike. A key characteristic of these coerced transfers is that they are not motivated by mutual gain, but instead by the gain of one party facilitated by superior force and institutional power. These transfers of resources and lives have shaped the course of history and have had important economic consequences and enduring legacies. But here we set aside the use of physical coercion and ask how societies organize the process of exchange motivated not by fear of physical harm but instead by the prospect of mutual gain. We also provide terms that allow us to evaluate some of these outcomes as better or worse than others. 4.1 Mutual gains from trade: Conflict and coordination In a modern economy we engage in indirect monetary exchange: selling some of our goods or some of our working time for money and using the money to purchase the goods we need rather than bartering directly as did the Libyans and Carthaginians. The principles of barter exchange, where goods are directly transferred among two parties without the use of money, however illustrate the fundamental considerations behind all types of exchange, including Figure 4.2: A statue of Herodotus. Considered by many to be the first historian, Herodotus lived in the fifth century BCE. Source: Wikimedia Commons. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S indirect monetary exchange. We will simplify by thinking about just two people who exchange goods directly with each other, thereby modifying the goods that they hold. To do this we will introduce two terms describing the bundles that each has before and after 173 R E M I N D E R As in Chapter 3 a bundle is just a list of the quantity of the goods (or other thing of value) that a person has. We refer to the bundles held by all of those involved in an exchange as an allocation. exchange: • The endowment bundle or endowment, the quantities of goods a person has before exchanging goods. • The post-exchange bundle the bundle a person has after exchanging goods with another person. The bundles held by each of the people (either before or after exchange) is called an allocation. Voluntary exchange: mutual gains and conflict over their distribution An exchange is voluntary if all parties to the exchange have the option to not engage in it but instead choose engage in the exchange. So each party must expect to be better off, or at least is no worse off, as a result of the exchange, which implies that each prefers (at least weakly) their post-exchange bundle to their endowment bundle. Recalling the meaning of a Pareto-comparison, we can see that if an exchange is voluntary for both parties, the post-exchange allocation must be a Pareto-improvement over the endowment, otherwise one or both of the parties would have refused to participate in the exchange. The stipulation that the A LLOCATION The bundles held by each of the people (either before or after exchange) is called an allocation. VOLUNTARY E XCHANGE An exchange is voluntary if all parties to the exchange have the option to not engage in it but instead choose engage in the exchange. So each party must expect to be better off, or at least be no worse off, as a result of the exchange, which implies that each prefers (at least weakly) their post-exchange bundle to their endowment bundle R E M I N D E R : An economic rent is the difference between a player’s fallback payoff and the payoff (profit or utility) they obtain from participating in an interaction. The gains from exchange from an interaction is the sum of the economic rents of all participants. in order for an exchange to be called voluntary, the post exchange allocation must be a Pareto-improvement over the endowment bundle is termed the voluntary transfer requirement. To make the idea of voluntary exchange concrete we often let the fallback position of the players be a bundle of goods that is their private property which they are free to dispose of in exchange or by gift to others, or to retain for themselves, excluding others. Let’s review some of the terminology from earlier chapters and explain how they are used to study the process of exchange. • A person’s fallback position is what they experience in the absence of the exchange and the utility number they assign to that bundle (that is, the utility of their endowment bundle which is considered to be her next best opportunity.) • The improvement in utility enjoyed by a party to an exchange is their rent resulting from the exchange, namely, the difference in utility associated with their post exchange bundle compared to their fallback position. P RIVATE PROPERTY Private property is the right to exclude others from the goods one owns, and to dispose of them by gift or sale to others who then become their owners. VOLUNTARY TRANSFER REQUIREMENT The stipulation that in order for an exchange to be called voluntary, the post-exchange allocation must be a Pareto-improvement over the endowment bundle is termed the voluntary transfer requirement. 174 MICROECONOMICS - DRAFT • The total rents received by parties to an exchange, also termed the gains from trade are the utilities of the exchanging parties at the outcome of the exchange minus the utilities at their fallback positions. . ?? The fact that an exchange is voluntary does not mean that is is fair. Some exchanges take place under conditions such that one party gains virtually all of the available rents. How the economic rents are divided between par- P RIVATE PROPERTY Private property is the right to exclude others from the goods one owns, and to dispose of them by gift or sale to others who then become their owners. ticipants is the distributional outcome of the exchange. The rents may be captured by one party, leaving the other with a different set of goods than her endowment but no better off. Or the rents may be split among the parties in a way that appears fair, or at least acceptable to both, as in the silent trade between the Carthaginians and the Libyans described by Herodotus. The division of the gains from exchange in the form of economic rents is parallel to the division of the pie in the Ultimatum Game of Chapter 2. D ISTRIBUTIONAL O UTCOME How the gains from exchange – the economic rents – are distributed between the people in an exchange; the share of the gains from exchange each player gets as a rent. Exchange therefore has two aspects: mutual benefit and conflict of interest: • Mutual benefit is possible because participants move from their endowment bundle to the post-exchange allocation where they share the gains from exchange and obtain an economic rent. • A conflict of interest is present because the gains from exchange can be divided in many ways among the parties who find themselves in conflict over who gets the larger share. Institutions and social norms govern the process of exchange that leads both to the re-allocation of goods, and to the distribution of the gains from trade. We will see that institutions and social norms have effects on: • Pareto efficiency, facilitating or obstructing the realization every opportunity for mutual gain among the parties to an exchange, and • The fairness of the distributional outcome, favoring one party or the other in the conflict of interest in the distribution of the economic rents. A major institutional challenge today is to find rules of the game that will have as the Nash equilibrium allocations that are both Pareto efficient and fair. We will return to the interplay of these two objective frequently in the pages that follow. 4.2 Feasible allocations: The Edgeworth box Lets consider a concrete setting in which two people might consider alternative possible distributions of two goods amongst themselves. Let’s say that H I S TO RY The Edgeworth box is named after the British economist Francis Ysidro Edgeworth (1845-1926) who is credited with having invented this clever way to represent exchange and bargaining. P R O P E RT Y, P OW E R , 8 7 B's coffee (kilograms), xB 6 5 4 3 2 Biko 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 z 0 1 2 Ayanda 3 4 5 6 7 A's coffee (kilograms), xA 8 9 10 10 9 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 8 7 B's coffee (kilograms), xB 6 5 4 3 2 1 Biko 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Better for Ayanda Better for Biko 0 1 2 Ayanda (a) An allocation, z Ayanda and Biko have to divide a total of 10 kilograms of coffee and 15 gigabytes of data between them. At the start, nobody owns the goods, the two quantities are simply amounts available to the two of them. Ayanda and Biko might now ask each other: what allocation of the coffee and data between the two of us would be the best? We use the notation x̄ = 10 and ȳ = 15 to stand for the total amount of coffee (x) and data (y) available. We define xA , and yA as the quantity of goods x (coffee) and y (data) in Ayanda’s bundle, and similarly xB and yB are the quantities in Biko’s. The amount of the two goods in their respective bundles can be anywhere from zero to the entire amount available, namely, x̄ and ȳ. Then, an allocation is a particular assignment of coffee and data to the two people that we can write as (xA , xB ; yA , yB ). An allocation is feasible if the amounts of coffee and data it gives to Ayanda and Biko is no greater than the amount available: xA + xB x̄ yA + yB ȳ Figure 4.3 (a) represents the total supply of the goods, with width and height equal to the total amount of coffee (x) and data (y) available. The box’s width is the total amount of x, x̄ (kilograms (kg) of coffee) and its height is the total amount of y, ȳ (gigabytes (gb) of data). We measure A’s allocation, (xA , yA ) from the lower left-hand corner of the box, and B’s allocation, (xB , yB ) from the upper right-hand corner. Any point in the box (or on its edges) is a bundle representing a feasible allocation of the two goods between the two parties, with the property that it fully exhausts the total supply of the two goods. You can see this because the width of the box is the total amount of x and the height of the box is the 3 4 5 6 7 A's coffee (kilograms), xA 8 175 9 B's data (gigabytes), yB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1 A's data (gigabytes), yA 9 B's data (gigabytes), yB A's data (gigabytes), yA 10 & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 10 (b) Indifference curves Figure 4.3: Feasible allocations that exhaust the supply of both goods. Figure 4.3a shows an example of a feasible allocation at point z. Figure 4.3b shows the direction in which each person prefers to move to increase their utility. When indifference curves are plotted in this rectangle the graph is called an Edgeworth box. 176 MICROECONOMICS - DRAFT total amount of y. Allocation z, for example, gives Ayanda 9 kilograms of coffee and 1 gigabyte of data and Biko 1 kilogram of coffee and 14 gigabytes of data (exhausting the 10 units of x and the 15 units of y). There are also many feasible allocations of the two goods that are not shown in the box. For example, if Ayanda and Biko each got 1 kilogram of coffee and one gigabyte of data, that would be feasible given the total amounts, but it could not be shown in the Edgeworth box because the Edgeworth box only shows allocations where the two people divide up all of the goods so that they sum to x̄ and ȳ. As we move to the northeast in the box, Ayanda gets more of both goods, and as we move to the southwest in the box, Biko gets more of both goods. Because both are self-regarding we show this on the figure with the arrows labeled: "Better for Ayanda" and "Better for Biko" respectively . How can we evaluate whether some allocations are better than others? To do this we can represent the preferences of the two parties by plotting their indifference curves in the box. This allows us to say for both Ayanda and Biko that for any two allocations (points in the box) that the first is preferred to the second, the second is preferred to the first, or the person is indifferent between the two. To do this we need to know the utility functions of the two. Both Ayanda and Biko enjoy consuming both coffee and data. Their utility functions are: Ayanda’s utility function uA (xA , yA ) Biko’s utility function uB (xB , yB ) We assume that the indifference curves for both parties exhibit decreasing marginal utility for both goods. To provide a concrete example, we will assume that both Ayanda’s and Biko’s utility functions are Cobb-Douglas, but in some cases that follow, with different preferences for coffee and data: A aA) B aB) Ayanda’s utility function uA (xA , yA ) = (xA )a (yA )(1 Biko’s utility function uB (xB , yB ) = (xB )a (yB )(1 In numerical examples we will often contrast two cases: • Identical: The two people have identical preferences for the two goods, such as a A = 12 , a B = 12 . • Different: The two people have different preferences, for example, such that A’s a A = 23 , whereas for B a B = 13 . So Ayanda has a stronger preference for coffee than Biko does. We can visualize the allocation of coffee and data between Ayanda and Biko using an Edgeworth box. An Edgeworth box allows us to see both people’s indifference curves in the same space to identify mutually beneficial trades. R E M I N D E R Recall that in Chapter 1, we used z to indicate the fallback position of people playing games in the general form of the Fishermen’s Dilemma game with ranked outcomes. At z, the people experience their B utilities, uA z and uz , as their utility at the fallback position, that is, their endowments if they do not trade. 15 15 14 14 13 13 12 12 uA3 10 z 11 10 A's data (gigabytes), yA uA2 B's data (gigabytes), yB uA1 11 ● & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 10 9 8 7 6 5 4 3 9 8 7 6 5 uB3 4 3 2 uB2 2 z 1 uB1 1 0 0 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 10 (a) Ayanda’s Indifference Curves 0 1 2 3 4 5 6 7 8 9 10 B's coffee (kilograms), xB Biko (b) Biko’s Indifference Curves Ayanda and Biko’s indifference curves are shown separately in Figure 4.4 panels a and b. In panel c. we plot the same indifference curves together in the Edgeworth box. Ayanda evaluates the allocations from the point of view of the lower left-hand corner, and her indifference curves represent higher utility 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 uA1 9 8 B's coffee (kilograms), xB 7 6 5 4 3 2 177 Biko 1 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 uA3 uA2 uB3 uB2 uB1 z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 B's data (gigabytes), yB A's data (gigabytes), yA P R O P E RT Y, P OW E R , 10 (c) Indifference curves in an Edgeworth Box Figure 4.4: Indifference curves and an Edgeworth box. In panels a. and b. we show three of Ayanda’s and Biko’s indifference curves respectively. In panel c., Biko’s indifference curves have been flipped 180°so that the origin in the lower left of panel b. has become the origin of the Edgeworth box at the upper right. as we move to the northeast in the box. Ayanda’s indifference map looks exactly the same in the Edgeworth box as it does in the separate plot, because in both cases the origin from which we measure her allocation is in the lower left-hand corner. Biko evaluates the allocations in the box, however, from the point of view of the upper right-hand corner, and his indifference curves represent higher utility as we move to the southwest in the box. It may help you understand how we superimposed Biko’s preferences on Ayanda’s if you think about what we called their "point of view." In panel’s a. and c., imagine Ayanda standing at the lower left origin and looking up her indifference map, as if the curves were contours of a mountain, the curves farther away being at higher altitudes. Now do the same with Biko, but for him when he looks to the north east in panel b., he is looking R E M I N D E R In Chapter 3 we defined the Cobb-Douglas (CD) family of utility functions as: u(x, y) = xa y(1 a) (with 0 a 1). The Cobb-Douglas utility function results in a marginal rate of y substitution, mrs(x, y) = (1 aa ) x . up his "utility mountain." But in panel c. he is standing at the upper left origin and the way up his utility map is to the south west. In the figures, at allocation z Ayanda and Biko have allocations (xzA , yA z) = (9, 1) and (xzB , yBz ) = (1, 14). The indifference curves that go through allocaA B B tion z provide Ayanda and Biko with utilities uA z = u2 and uz = u2 . A In panels a. and c., uA 2 = uz is Ayanda’s indifference curve through z. In panB els b. and c., uB 2 = uz is Biko’s indifference curve through z. The indifference maps for both Ayanda and Biko have indifference curves through every point in the box, but (following "the map is not the territory" principle) we show only three in the figure. M-Note 4.1: Evaluating utilities at an allocation Given the assumption of that Ayanda’s utility function is a Cobb-Douglas with a A = 23 and Biko’s utility function is a Cobb-Douglas with a B = 13 , we can calculate their utilities at the allocation z. Remember that the exponents in the Cobb-Douglas utility function represent M - C H E C K Biko’s indifference map would look exactly the same as in Figure 4.4 b. if we rotated the Edgeworth box 180°to measure Biko’s allocation from the lower left-hand corner. 178 MICROECONOMICS - DRAFT the person’s intensity of preference for the good. In this example, Ayanda likes coffee more than Biko does. 2 1 Ayanda has a Cobb-Douglas utility function uA (xA , yA ) = (xA ) 3 (yA ) 3 : • She has 9 kilograms (kgs) of coffee and 1 gigabyte (gb) of data. • So her allocation at point z is (xzA , yA z ) = (9, 1) • At her allocation z her utility is uA (xzA , yA z ). 2 1 • So for 9 kgs of coffee and 1 gb of data: uA (9, 1) = (9) 3 (1) 3 = 4.326749. Checkpoint 4.1: Biko’s utility at allocation z Using the method shown in M-Note 4.1, what is Biko’ utility at the allocation given by point z in the Edgeworth box. 4.3 The Pareto-efficient set of feasible allocations Which allocations in the Edgeworth box are Pareto-efficient? It’s easy to see that simply throwing away some of x or y cannot be efficient because allocating those portions to Ayanda and or Biko instead would have made at least one of them better off without making the other worse off. So Pareto-efficiency also requires that Equations 4.1 are satisfied as equalities, not as inequalities. By construction, any of the great many allocations in the Edgeworth box allocates all of the coffee and data to one or the other participant, and meets this criterion. To narrow things down, Ayanda and Biko could agree that the final allocation chosen must be Pareto-efficient. In Figure 4.5 we show Ayanda and Biko’s indifference curves through an arbitrary allocation z and three more indifference curves for each person: two indifference curves higher and one indifference curve lower than for allocation z. The endowment allocation is not Pareto-efficient Think about z as a hypothetical allocation, for example, if Biko said: "Ayanda, how about you have 9 kg of coffee and I get the 1 kg remaining, while I get 14 gb of the data, and you get the 1 gb remaining." We can see, however, that z in Figure 4.5 is not Pareto efficient. The reason is that at the allocations given by point z, Ayanda’s and Biko’s indifference curves: • intersect, which means • they have different slopes, • indicating different marginal rates of substitution R E M I N D E R In games like the Ultimatum Game in Chapter 2 any allocation of the pie in which the entire endowment is allocated to one of the players or the other – in other words " no money left on the table" is Pareto-efficient. But the allocations resulting from the Ultimatum Game are frequently inefficient because when the Responder rejects the Proposers offer both players get zero, and all of the money is left on the table. M - C H E C K Even if for some reason we were not to allow the allocation to involve fractional quantities of the goods and require that allocations be integers, there 176 possible allocations to be exact (that’s 11 ⇥ 16, in case you are wondering, because we would then have to include zeroes as possible allocations for the goods). R E M I N D E R The marginal rate of substitution is the negative of the slope of the indifference curve. It is also equal to the ratio of the marginal utilities of the two goods, x and y, i.e. mrsA (x, y) = uA x . The marginal rate uA y of substitution is also the willingness to pay for x in terms of y. The people’s marginal rates of substitution have the dimensions data/coffee (data for coffee). R E M I N D E R For an outcome to be Paretosuperior to another, at least one participant must be made better off – get higher utility – and no participant can be made worse off – get lower utility. P R O P E RT Y, P OW E R , 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 8 uA1 B's coffee (kilograms), xB 7 6 5 4 3 2 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Pareto− efficient curve uA3 uAz uB1 uBz d uB3 tA uB4 i tB 0 Ayanda 1 2 3 4 5 h 6 7 A's coffee (kilograms), xA z 8 9 Figure 4.5: Pareto-efficient allocations To 2 10 goods, which means their willingness to pay to acquire more of one or the other good differ; • and this means that there is a feasible Pareto-improving exchange that has not been realized, • so these allocations are not Pareto efficient. Figure 4.5 provides us with a numerical demonstration of the above logic. We have seen that the difference between the two people’s marginal rates of substitution at the point z indicates that they can make a Pareto-improving trade – Ayanda giving up some of her coffee in return for some of Biko’s data – at their endowment allocation z. In M-Note 4.2 we show that Ayanda’s mrs is 29 and Biko’s is 7 in Figure 4.5. This means that Ayanda is willing to pay at most a kilogram of coffee for 29 of a gigabyte of data, while Biko is willing to trade at most 7 gb of data for one kg of coffee. A mutually beneficial trade could therefore take place at any price of coffee between 29 of a gb of data and 7 gb of data. The low price would benefit Biko, with Ayanda not improving her utility at all. Correspondingly, if they traded at the high price Ayanda would make all the gains. We return to how the price might be determined later. For now, we can eliminate point z in Figure 4.5 as a candidate for being a Pareto-efficient alloca- 1 make this figure we let uA = (xA ) 3 (yA ) 3 and • meaning that Ayanda and Biko have different offer-prices for the two tion. 179 Biko 1 B's data (gigabytes), yB A's data (gigabytes), yA 10 & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S uB = 1 ( xB ) 3 2 ( yB ) 3 . So Ayanda prefers has a stronger preference for coffee, and Biko for data (they have asymmetrical preferences). Allocation h is Pareto-superior to allocation z, but it is not Pareto-efficient because an alternative point, e.g. allocation tB , is Pareto-superior to point h (Biko is better off without Ayanda being worse off). All points along the Pareto-efficient curve between i and tB would be both Pareto-superior to h and z and Pareto-efficient. 180 MICROECONOMICS - DRAFT M-Note 4.2: mrs in the Edgeworth box At allocation z (9, 1), (1, 14) in Figure 4.5, we can calculate each person’s marginal rate of substitution and compare them. We computed what a person’s mrs(x, y) is when she has Cobb-Douglas utility in M-Note 3.4 in Chapter 3. We obtain Biko’s from the same reasoning. We shall assume for this example that the two have asymmetrical preferences as in Figure ??. Let’s start with Ayanda, assuming a A = 23 : uA yA • mrsA (x, y) = uxA = 2 xA y 1 2 A A • Substitute in A’s allocation at z: mrsA z (xz , yz ) = 2 9 = 9 Ayanda is willing to pay 29 of a gigabyte to get a kilogram of coffee, or to sell a kilogram of coffee for 29 of a gigabyte of data. Now for Biko, assuming a B = 13 : uB yB • mrsB (xB , yB ) = uxB = 12 xB y 1 14 B B • Substitute in B’s allocation at z: mrsB e (xe , ye ) = 2 1 = 7 Biko is willing to pay 7 gigabytes of data for a kilogram of coffee, or to sell kilogram of coffee for 7 gigabytes of data. We can see that mrsA < mrsB because 29 < 7. This shows up in Figure 4.5 where the slope of Ayanda’s indifference curve is steeper than the slope of Biko’s indifference curve at allocation z. Which allocations are Pareto efficient? The mrsA = mrsB rule The same reasoning allows us to eliminate most of the other points too. Remember the demonstration that showed point z to be Pareto-inefficient started with "at the allocations given by these points Ayanda’s and Biko’s indifference curves intersect." We explain this further in M-Note 4.2. So any allocation at which the indifference curves intersect, like point h in Figure 4.5 cannot be Pareto efficient. To find the Pareto-efficient allocations, we need to determine which allocations remain after we have eliminated all of those at which the indifference curves cross. To do this we can run the above reasoning in reverse. If the two indifference curves (one of Ayanda’s, one of Biko’s) share a common point (that is, that represent the utilities at a particular allocation) but do not intersect, then the two indifference curves must be tangent. This tells us (reversing the logic above about indifference curves that intersect) that if Ayanda’s and Biko’s indifference curves: • are tangent, this means that • they have the same slopes, indicating P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 181 • identical marginal rates of substitution, • meaning that Ayanda and Biko have the same willingness to pay for the two goods. • This is the same as saying that their maximum willingness to pay to acquire more of the other’s good is not greater than the least price at which the other would part with their good • and this means that there is no feasible Pareto-improving exchange into which both would voluntarily enter • so the status quo allocation is Pareto efficient. This gives us the following rule for an allocation between two players, A and B, being Pareto efficient: The mrsA = mrsB rule: mrsA (xA , yA ) = mrsB (xB , yB ) (4.1) This rule differs from the seemingly similar mrs = mrt rule for a single individual because this new rule applies to strategic interactions among two or more inter-dependent actors, of the kind that occur in markets for labor, credit, and R E M I N D E R : T H E mrs = mrt RU L E We derived a similar rule for single person interactions in Chapter 3 The mrs = mrt rule (with a few exceptions) identifies the constrained optimal allocation for a single individual as the bundle at which the marginal rate of substitution (the person’s willingness to pay for more of the y-good) is equal to the marginal rate of transformation (the opportunity cost of getting more of the y-good). many goods. The superscripts A and B are there to remind you that two (or more) players are involved in this rule. The points tA , tB and i lie on the purple Pareto-efficient curve in Figure 4.5. We will often abbreviate the Pareto-efficient curve to PEC. The Pareto-efficient curve consists of all Pareto-efficient allocations, including Ayanda getting all of both goods, or the reverse. A M - C H E C K Like the mrs = mrt rule, mrs = mrsB does not work in every case. The Confining allocations to the Pareto-efficient curve limits the choices that Pareto-efficient point may be a corner solution (not a tangency) at which one of the goods is not consumed at all by one of the players, and a tangency identified by the rule may be a minimum not a maximum. The reasons are the same as were explained for the mrs = mrt rule. Ayanda and Biko need to make. But the question is still far from answered. Moving from one Pareto-efficient allocation to another must make one of the participants better off and the other worse off. The Pareto efficiency criterion is not going to help them decide which of the points on the Pareto-efficient curve they would consider to be the best. So they face a problem and a conflict of interest. • The problem is that there are still innumerable Pareto-efficient outcomes on the PEC and they need some way to decide which one to choose. • The conflict of interest is that Ayanda prefers points on the PEC to the northeast in the Edgeworth box, while Biko prefers points to the southwest, so they will not agree on which Pareto-efficient division of the coffee and data to make. M-Note 4.3: Computing the Pareto-efficient Curve We will use mrsA = mrsB rule to work out the equation for the Pareto-efficient curve. PARETO - EFFICIENT CURVE The Paretoefficient curve is all outcomes that are Pareto-efficient. At a Pareto-efficient outcome the marginal rates of substitution of the two parties are equal so the mrsA = mrsB holds. The Pareto efficient curve is sometimes called the "contract curve", a term we do not use because there need not be any contract involved (e.g. when an outcome in our thought experiment was implemented by the Impartial Spectator. 182 MICROECONOMICS - DRAFT To find the Pareto-efficient curve, we set Ayanda’s marginal rate of substitution equal to Biko’s marginal rate of substitution. We already know that mrsA (xA , yA ) = B A 2 xyA and mrsB (xB , yB ) = 12 xyB . We also know that x̄ = xA + xB = 10, so xB = x̄ xA and ȳ = yA + yB = 15, so yB = ȳ yA . Solutions to these equations for xA , yA , xB , yB are Pareto-efficient allocations. We set the marginal rates of substitution equal to each other and use these conditions to find the Pareto-efficient curve: mrsA (xA , yA ) = mrsB (xB , yB ) = 4(10 xA yA 4 A x xA )yA = 1 ȳ yA 2 x̄ xA 15 yA 10 xA xA (15 yA ) 40yA 4xA yA = 15xA (40 3xA )yA = 15xA yA = 15xA 40 3xA 2 Pareto-efficient Curve yA = xA yA The Pareto-efficient allocations lie on an upward sloping curve between the two origins. Checkpoint 4.2: Conflict and symmetry of preferences on the Paretoefficient curve a. Using Figure 4.5 do the following: i. Explain Ayanda’s and Biko’s preference among the Pareto-efficient points tA , tB , and i. ii. Show that they rank these points in opposite order. iii. Explain why for any two points on the Pareto-efficient curve, Ayanda will prefer one point and Biko another point; they will never agree on which is preferable. b. Work out the formula for the Pareto-efficient curve when the two people have identical Cobb-Douglas utility functions where a A = a B = 12 . (It’s easier than the asymmetrical case.) The solution is that the Pareto-efficient curve is given by yA = 32 xA . But it’s important for you to work out how to get the solution. 4.4 Adam Smith’s Impartial Spectator suggests a fair outcome Ayanda and Biko are going to have to figure out some way – other than each simply trying to get more – for picking an allocation. This means stepping back and looking at the problem without thinking about their own particular preferences. They would probably experiment with some simple rules. They could adopt: • "finders keepers" rule and allocate the goods to whoever had first discovered the discarded coffee and data; but this might not seem fair. • the fifty-fifty norm of the landlords and sharecroppers in Chapter 2, and E X A M P L E To see how maximizing total utility might lead to unacceptable outcomes, think about two people, one who in order to minimize her carbon footprint or for other ethical reasons has cultivated a simple life style and is not much interested in increasing her material consumption and the other who has cultivated a taste for luxuries and will be miserable without them. Maximizing total utility would require giving most of the goods to the second person. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S each take half the quantity of the two goods; but if they have different preferences (as is the case in panel b of Figure 4.5) splitting both goods equally would not even be Pareto-efficient (an equal split is not on the purple Pareto-efficient curve.) • the maximize total utility principle; but this places no value on equality, and might result in selecting an allocation in which one person had most of the goods (and utility) and the other little of either. To develop more satisfactory rules, they might consult an Impartial Spectator a fair and impartial spectator who can assist them (and us) in reasoning about what a good outcome might be. We use upper case letters for her name to remind you that she is an entirely made-up character, a thought experiment, and not a part of the game in which Ayanda and Biko are engaged. The Impartial Spectator is not a person, she is a thought experiment representing our conscience, allowing us to explore differing values and how they would lead us (and Ayanda and Biko) to select a particular allocation as the best. We’re going to follow the Impartial Spectator’s thinking by looking at different criteria that she could adopt. For example, she could ask: • Are the procedures that determined the allocation fair? • Is the outcome itself fair? The first criterion is referred to as a procedural judgement, and therefore she judges the outcome based on the procedure used to acquire the goods. She would ask for example if the original endowment bundles had been acquired fairly, for example through hard work, gifts, or exchanges in which both Ayanda and Biko had an equal opportunity to acquire the goods. She would go on to inquire if the process of trading had itself been fair: for example did either of them have unfair advantages in determining the price at which they would exchange. The second criterion is called substantive: it asks about the substance of the resulting allocation, asking for example if it is fair (no matter how it came about). Both criteria are important, but we will focus on the substantive judgements because it allows us to illustrate how the Impartial Spectator could select the “best" allocation by solving a constrained optimization problem. For the Impartial Spectator to make judgments among Pareto-efficient allocations that give Ayanda and Biko different levels of utility using the constrained optimization method, she needs to refer to two pieces of information: • The set of all Pareto efficient combinations of utility levels that Ayanda and Biko could experience by allocating the goods in different ways; 183 184 MICROECONOMICS - DRAFT • How she (the spectator herself) values each of these combinations of the utility levels of the two. The utility possibilities frontier Setting aside Pareto-dominated allocations, the Impartial spectator will concentrate on the boundary of the set of feasible utility pairs of the two. This is called the utility possibilities frontier (UPF) and it, shows all combinations of Ayanda and Biko’s utilities associated with allocations on the Pareto-efficient curve. In Figure 4.6 In panel a we show an Edgeworth box of the two player’s allocation problem in which they have identical preferences in the way they each value coffee and data. In panel b, we show the utility possibilities frontier for this case. For the moment, ignore the downward-sloping blue lines. The utility possibilities frontier shows Ayanda’s utility (uA ) on the horizontal axis and Biko’s utility (uB ) on the vertical axis as we move from one extreme of the Pareto-efficient curve in figure 4.5 to the other. The UPF is downward-sloping because the participants are in conflict over who gets what share of the possible distributions of utility as the allocation changes on the Pareto-efficient curve in the Edgeworth box. The UPF is constructed from the Pareto-efficient curve by translating each Pareto-efficient allocation (xA , yA ; xB , yB ) into a point (uA (xA , yA ), uB (xB , yB )) that represents the utility levels of the two participants at that allocation. To construct it, take any point on the Pareto-efficient curve in Figure 4.5, say point tA , then read U TILITY P OSSIBILITIES F RONTIER (UPF) The utility possibilities frontier is a curve plotted with uA on the horizontal axis and uB on the vertical axis that shows the utility of the two participants at all Pareto-efficient outcomes. from the two indifference curves through tA the two levels of utility of Ayanda and Biko at that allocation (namely 8.52 and 3.74 respectively), then go to Figure 4.6 where those two utility levels become the coordinates in the utility possibility graph of point tA in the Edgeworth box graph. Points tA , i, and tB correspond to the same lettered points in Figure 4.5 and portray the utilities of each of the two at these Pareto-efficient allocations. In similar fashion, points z and h correspond to the same letters in Figure 4.5, but these allocations, being Pareto-inefficient are of no interest to the Impartial Spectator. As in the case of other feasible frontiers the negative of the slope of utility possibility frontier, DuB , DuA is the marginal rate of transformation of B’s utility into A’s utility by progressively giving A more of the goods and B less. This is also the opportunity cost of A having more utility in terms of the sacrifice in B’s utility necessary to allow this. A steep utility possibility frontier means that for A to gain one unit of utility, B must sacrifice a lot. R E M I N D E R The utility possibilities frontier is similar to what we did in Chapter 1 to understand the Pareto efficiency of different game outcomes. The UPF is another feasible frontier introduced in Chapter 3, since it shows the feasible combinations of utility possible given the available goods and the preferences of the participants. P R O P E RT Y, P OW E R , B's coffee (kilograms), xB 8 7 6 5 4 3 2 Biko 1 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Pareto−efficient curve d uA3 uA1 uAz A t i tB uB3 uB4 uBz uB1 h 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA z 8 9 10 (a) Identical Preferences Edgeworth box 13 185 Infeasible combinations of utility 12 11 10 tB 9 B's Utility, uB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 B's data (gigabytes), yB A's data (gigabytes), yA 10 & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 8 mrs = mrt Impartial Spectator's iso−social welfare curves 7 i h 6 5 4 3 w6 tA z w5 w4 w3 w2 w1 Feasible combinations of utility 2 1 Utility possibilities frontier 0 0 1 2 3 4 5 6 7 8 A's Utility, uA 9 10 11 12 13 (b) The utility possibilities frontier (UPF) and the Impartial Spectator’s iso-social welfare curves (w) Figure 4.6: The utility possibilities frontier Checkpoint 4.3: The UPF and the PEC 1. Explain why the utility possibilities frontier in Figure 4.6 is downward-sloping. 2. Explain why if the utility functions of the two differ, an even split of the two goods – half of each to Ayanda and half of each to Biko – could not be the Impartial Spectator’s choice of the best allocation. The Impartial Spectator’s social welfare function (UPF) and the Impartial Spectator’s iso-social welfare curves (w). The utility functions of the two players used to create panel a are identical, with in both cases a = 0.5. Because they both value the two goods in the same way, they consume them in the same proportions at all points on the PEC. The only differences is which player has more. Each point in Panel b. corresponds to an allocation in the Edgeworth box shown in panel a. The downward-sloping curves in Figure b are the Impartial Spectator’s iso-social welfare curves, corresponding to six levels in his judgement of social welfare w1 through w6 . Social welfare is lower at points closer to the origin. The allocation given by point i is the social optimum determined by the mrs = mrt rule. Which point on the UPF – in other words which allocation of the goods between Ayanda and Biko – the Impartial Spectator ranks as best will depend on her values. She has to compare how much she values the utility of Ayanda and Biko respectively and how this varies depending on the level of utility that each are experiencing. To do this she has to be able to compare how much the levels of utility for the two for each of the allocations on the UPF. She knows that Ayanda prefers allocation tA to tB (and that Biko ranks these two allocations the other way around). She needs to treat the utility of each like ordinary numbers that measure the size not just the rank of something, in this case the cardinal utility of each. A summary of the Impartial Spectator’s evaluation of different utility distributions (uA , uB ) is provided by her social welfare function, W (uA , uB ). This is similar to the utility function that expresses a person’s preferences over bundles of goods, (x, y), but remember the Impartial Spectator is not a person, but a thought experiment. This is why it is called the social welfare function rather than the Impartial Spectator’s utility function. A social welfare function provides a way of treating the utilities of the citizens as being cardinal S OCIAL W ELFARE F UNCTION A social welfare function is a representation of "the common good" based on a some weighting of the utilities (uA , uB , and so on) of the people making up the society. We can write a social welfare function in the form W (uA , uB ). 186 MICROECONOMICS - DRAFT numbers that are comparable across people (like height or weight). R E M I N D E R Assigning cardinal utility numbers to bundles means that we can make statements like: An example is a social welfare function that expresses total welfare as the • for Annette, the outcome (x0 , y0 ) is twice as good as (x, y) but also product of the utility of the citizens, each utility raised to some exponent. Example Social Welfare Function: W (uA , uB ) = (uA )l (uB )1 l (4.2) This social welfare function has the same form as a Cobb-Douglas utility function: the participants’ levels of utility are the "goods" for the Impartial Spectator. When l = 0.5 = 1 l , then the Impartial Spectator: • the sum of the utility experienced by Annette and Brenda is greater with outcome (x0 , y0 ) than with outcome (x, y) because uA (x0 , y0 ) + uB (x0 , y0 ) > uA (x, y) + uB (x, y) The sum of the utilities of the two – in the second statement – is an example of a social welfare function. • weights the two peoples’ utilities equally ; and • places diminishing marginal value on increases in the utility of either of Ayanda or Biko; the more they consume of the goods the greater is their utility and therefore the less they add to the Impartial Spectator’s social welfare. Because the Impartial Spectator values the two people’s utilities equally, and (in the judgement of the Spectator) the marginal value of increased utility is diminishing, she will not rank highly any outcome in which one or the other gains most of both goods. Just as we can use indifference curves to represent a person’s utility function over goods, we can use iso-social welfare curves to represent the Impartial Spectator’s social welfare function over the utility distribution between people. The level of social welfare is the same along an iso-social welfare curve, just as utility was the same along an indifference curve. Given the Impartial Spectator’s social welfare function, the problem of choosing the Pareto-efficient allocation of coffee and data becomes a constrained maximization problem similar to those we studied in Chapter 3. The feasible frontier for the Impartial Spectator is the utility possibility frontier, because it represents the levels of utility that are achievable given the amount of goods available and the preferences of the participants. The iso-social welfare curves of the social welfare function are analogous to indifference curves for a single individual, but apply to the utilities of the two people not the two goods consumed by the single individual and express the valuations of the Impartial Spectator, not the preferences of the individual. Similar to the individual indifference curve, the negative of the slope of the isosocial welfare curve at any point (uA , uB ) is the Impartial Spectator’s marginal rate of substitution of Ayanda’s utility in terms of Biko’s utility. And we can use the mrs = mrt rule to find the constrained social welfare-maximizing allocation. It is the point where the UPF is tangent to an iso-social welfare curve. I SO - SOCIAL WELFARE CURVE Iso-social welfare curves show constant or equal ("iso") levels of welfare, W̄ , for different combinations of utility between A and B. The negative of the slope of the iso-social welfare curve is the Impartial Spectator’s marginal rate of substitution (mrsSW (uB , uA )) of Ayanda’s utility for Biko’s utility. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S For example, suppose the Impartial Spectator’s social welfare function is: 1 1 W (uA , uB ) = (uA ) 2 (uB ) 2 which puts a identical weight on the utilities of the two parties, then given that Ayanda and Biko have identical utility functions, the social welfare maximum shown in the Edgeworth box in Figure 4.5 Panel a is xA = 5, yA = 7.5, xB = 5, yB = 7.5 or a fifty-fifty split of each good. In the utility possibility frontier graph in Figure 4.6 this is point i. If their preferences differed, then the social optimum would result in each getting different amounts of x and y. Different Impartial Spectators might have different social welfare functions to rank distributions of the utilities of the two parties, leading to the choice of different social welfare maximizing Pareto-efficient allocations. Societies do not have an Impartial Spectator to determine how to weight the competing interests of society’s members in a social welfare function. Instead, in a democratic society we debate the question of distribution and sometimes come to a consensus (and sometimes to a deadlock). Controversy about the rights and wrongs of economic policies such as the tax rates paid by wealthy people and the provision of public services to all, are often implicitly about the weights (such as a, in Equation (4.2)) that policy-makers should place on the well-being of different people. Here we see a sharp contrast between the Pareto-efficiency criterion and the maximization of social welfare. Preferring a particular Pareto-efficient allocation over an alternative allocating in which both are worse off cannot be a matter of conflict. Maximization of some particular social welfare function subject to the constrain of the utility possibility frontier – some gaining and some losing depending on the social welfare function used – and is certain to be controversial. The imaginary Impartial Spectator helps us understand how values dictate what we think of as better or worse allocations. These outcomes, as we have seen in previous chapters and we will now see in greater detail, depend on the rules of the game. So the Impartial Spectator will have something to say about how we evaluate which are better or worse institutions by which organize the process of exchange. 4.5 Property rights and participation constraints The scenario of Ayanda and Biko enjoying their coffee and data in their student residence and deciding how to allocate them helps us understand the abstract issues of Pareto efficiency and fairness. Very similar issues arise when instead we consider Ayanda and Biko to be total strangers, interacting in a market. But in this new setting the allocation will not be determined by some imaginary Impartial Spectator. Instead, the allocation will be determined by 187 188 MICROECONOMICS - DRAFT who initially owns which goods and the rules of the game that regulate how Ayanda and Biko might benefit by exchanging some of their goods with each other. Market institutions: Property rights and participation constraints Nobody actually owned the data and coffee that the Impartial Spectator allocated in our thought experiment and Ayanda and Biko were not really engaged in a game. This is not how markets work. Key aspects of the rules of the market game are: • The rule of law establishes that the institutions – the laws and other informal rules – governing the interaction are observed, and not violated by arbitrary acts (for example theft of the others goods by one of the traders or confiscating by a government official). • Private ownership. At any moment in the game the goods are the private property of one or the other of the players, so a point in the Edgeworth box indicates a distribution of property between the two. The distribution at the start of the game is called each player’s endowment. • Private property and the rule of law mean that each player as option to refuse offers so any exchanges that a player will agree to participate in must be Pareto improvements over the endowment • Asymmetric bargaining power will affect the nature of the exchanges that are executed, and who captures the greater share of the gains from exchange. Private property does not distinguish between the two parties: each have identical rights to exclude the other from their bundle of goods. This would be true even if Biko initially owned all of the goods and Ayanda had none or the other way around. In this respect private property rights provide a level playing field because the right to exclude others from the use of your goods does not depend on how many goods you have, or on your identity. The exchange process begins with the property people "start with," that is, their endowment allocation. These endowments exist before the exchange we are considering happens. But we are cutting into time at a particular mo- H I S TO RY It has not always been true that one’s property rights did not depend on your identity. In many societies, some people – such as women – did not have the right to own property, and some people – such as enslaved people – were treated as property. ment. These endowments, which are the status quo of our game, are the result of similar games played in the past, and also other games in which who owns what goods may have been determined by force and not by voluntary exchange. This means that unlike the Impartial Spectator starting with a clean slate – any allocation in the Edgeworth box is up for consideration – and advising Ayanda and Biko on the division of a pile of goods they have tripped over in their student residence, market exchange starts from one particular point in the Edgeworth box the endowment allocation. The rules of the E NDOWMENT ALLOCATION The ownership of goods at the start of the game is termed the endowment allocation. It is the starting point of the game, but in applications to real economies who owns what at any point in time is the outcome (not the starting point) of other interactions that have determine who owns what. P R O P E RT Y, P OW E R , B's coffee (kilograms), xB 8 7 6 5 4 3 2 Biko 1 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Pareto−efficient curve d uA3 = 8.52 uA1 uAz Pareto−improving lens tA i tB uB3 uB4 uBz = 3.74 uB1 h 0 Ayanda 1 2 3 4 5 6 7 A z 8 9 13 189 A's participation constraint, uAz 12 11 10 tB 9 B's Utility, uB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 B's data (gigabytes), yB A's data (gigabytes), yA 10 & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 8 7 i 6 Bargaining set 5 4 B's participation constraint, uBz tA z 3 2 Utility possibilities frontier 1 0 10 0 1 2 3 A's coffee (kilograms), x (a) Edgeworth box for self-regarding Ayanda and Biko 4 5 6 7 A's Utility, uA 8 9 10 11 12 13 (b) The utility possibilities frontier game then determine how the two can move to some other post exchange allocation. The endowment allocation is important for two reasons: • it is the starting point of the process and • and also, because the exchange is voluntary meaning they can refuse to trade, it is their fallback position, that is, the worst they can do. Figure 4.7: Edgeworth box, the utility possibility frontier, and the bargaining set. In Panel a uAz is Ayanda’s utility at her endowment and is her participation constraint (shown by indifference curve uAz ) and uBz is Biko’s utility at his endowment and is his participation constraint (shown by indifference curve uBz ).In Panel b) the coordinates of the x- and y-axis intercepts of the utility possibilities frontier give the utilities of the two when either Ayanda ( x-axis intercept) or Biko (y-axis intercept) have all of the economic rents. The participation constraint (PC) To see how the second bullet above will narrow down that the post-exchange allocation can be, starting at any given endowment allocation we introduce the following notation, along with panel a. of Figure 4.7 (we will explain panel b. below). The endowment bundle of person i is (xzi , yiz ) where the superscript indicates who person i is (i = A for Ayanda, i = B for Biko). This allocation is point z in the figure. It is identical to point z in previous figures, but instead of being some hypothetical allocation that the Impartial Spectator was trying out it is now something entirely different: it is what Ayanda and Biko own at the start of the game. From point z in the figure you can see that Ayanda’s and Biko’s endowments of coffee and data are: • Ayanda’s endowment: (xzA , yA z ) = (9, 1) • Biko’s endowment: (xzB , yB z ) = (1, 14) Introducing history in the form of initially privately owned endowments, along with the voluntary transfer requirement, limits the possible allocations that can result from exchange. Because they can refuse any deal and therefore experience the utility from R E M I N D E R The participation constraint is also the fallback in the exchange scenario, the utility that a person can certainly secure if they choose not to participate in exchange at all. 190 MICROECONOMICS - DRAFT their endowment bundle, they will not accept any post-exchange bundle that makes them worse off than their fallback utilities. The indifference curves, uA z and uB z , that include the endowment point are the post exchange bundles that yield a utility identical to their fallback position. These two indifference curves are called their participation constraints. They are called participation constraints because Ayanda will not participate in (that is she will refuse) any offer that would give her a post exchange bundle below and to the left of uA z . Likewise Biko will not participate in any offer that would give him a post exchange bundle above and to the right of uB z (labeled as uB z = 3.74 in Figure 4.7). The right to refuse an exchange that will make a player worse off – the basis of the participation constraints – reduces the possible set of post exchange allocations consistent with voluntary exchange, and starting from the endowment allocation indicated by point z in panel a. The yellow colored space between the two constraints – the indifference B curves, uA z and uz – is the entire set of allocations that are Pareto superior to point z and which therefore could be the result voluntary modifications of the endowment allocation by means of exchange. This area is called the Pareto-improving lens. 4.6 Symmetrical exchange: Trading into the Pareto-improving lens PARETO IMPROVING LENSThe set of allocations that are Pareto superior to the fallback options of the players is the Pareto improving lens – shaded yellow in the figures to follow in the rest of the book. In this section we start with the assumption that the two traders have identical preferences. That is, that their Cob-Douglas utility functions have a A = a B = 0.5. We used a hypothetical point z in Figure 4.5 to show that an allocation where the indifference curves cross cannot be Pareto-efficient. Our demonstration consisted of showing that at such an allocation both Ayanda and Biko could benefit from exchange. We can now use the same reasoning to illustrate how starting at point z, now an endowment allocation – a real distribution of ownership of two bundles – the two could trade into the Pareto improving space, and eventually all the way to the Pareto-efficient curve. Each person has a willingness to pay for x in terms of y, their marginal rate of substitution at the endowment allocation z. Ayanda’s maximum willingness to pay is her mrsA (9, 1) = 19 and Biko’s maximum willingness to pay is his mrsB (1, 14) = 14. The difference between Ayanda and Biko’s willingness to pay (mrs) signals an opportunity for Ayanda to trade data with Biko at a rate of exchange between her own marginal rate of substitution and Biko’s marginal rate of substitution. A small exchange on these terms would move them to a post-exchange allocation upward and to the left of the endowment. With a A = a b = 0.5, their marginal rates of substitution are mrsA = M-CHECK yA xA 1 9 and mrsB = yB xB = 14. This means they could make an exchange at a “price” between 19 of a gigabyte for a kilogram of coffee (Ayanda’s mrs) and 14 gigabytes of data for a kilogram of coffee (Biko’s mrs). = P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S To stress that the game is entirely symmetrical imagine that they have agreed on a set of rules to determine the price and the amounts to be exchanged. At any allocation at which the mrs of the two differs (meaning their indifference curves intersect), take the following steps: 1. Pick a "price" midway between the mrs of the two. (This means that at point z the price would be 14 + 19 divided by 2, or 7.06.) 2. Ask the amounts that each would like to transact at the price of 7.06 gb of data for a kilo of coffee, for example how much coffee Ayanda would like to ’sell’ at this price, and how much coffee Biko would want to ’buy’(these desired amounts will differ between the two); 3. Because the transfer has to be voluntary (nobody can be forced to buy more than they wish), transfer the amounts desired by the person who wishes to transact least. 4. At the resulting post-exchange allocation determine if the indifference curves are intersecting. If so return to step one and continue. If not (that is, if the indifference curves are tangent) end the game with this final allocation. We can see that by this process the two will have moved, step-by-step from the endowment allocation at point z to a final post-exchange allocation that will be on the Pareto-efficient curve. We know that they will get there for two reasons: • Trades are Pareto-improving: each trade they take moves them in the direction of the Pareto-efficient curve because moving in the other direction could not be a Pareto-improvement and would violate the voluntary transaction requirement. • Trade concludes at a Pareto-efficient outcome: by the rules of the game they have adopted they will keep on exchanging until they are at a place where their mrs’s are identical, which must be on the Pareto-efficient curve. They could have adopted a different set of rules or institution for exchange. For example, they could have said that for step 1 above there will be two alternative prices, one just a little less than Biko’s willingess to pay, and the other just a little more than the lowest price at which Ayanda would part with her coffee; and then just flipped a coin to see which of these prices they would use in that transaction. Having made that transaction, do another coin flip to see whose preferred price will be used, and so on until they reached a Pareto efficient point at which no further trade was possible. Other than knowing that they would eventually get to the Pareto-efficient curve, we do not know which specific point on the curve they would get to. If the coin flips went in favor of Ayanda, they could end up close to tA with Biko 191 192 MICROECONOMICS - DRAFT sharing very little of the gains from exchange. Or it could have gone the other way, somewhere near point tB . They even could have ended up at point i the allocation chosen by the Impartial Spectator. The utility possibilities frontier in Panel b. of Figure 4.7 translates these allocations and the transactions supporting them into the utilities of the two players. The Pareto-improving lens in panel a. corresponds to the bargaining set in panel b. The first panels shows all of the allocations – denominated in quantities of x and y allocated to the two – that are Pareto improvements over the endowment allocation. The second – the bargaining set – shows the utility levels associated with every allocation in the Pareto-improving set. The yellow area in panel b. is called the bargaining set because it compares outcomes of their bargains relative to their fallback position (point z). The bargaining set shows all of the possible distributions of the rents (utility greater than their fallback positions) that might result from their bargaining, depending on the rules governing how they bargain. These rules will determine the extent of the bargaining power of the players. Checkpoint 4.4: Pareto improvements, rents, and Pareto efficiency If point h is the post exchange allocation based on the endowment allocation of point z, explain the following: a. Did Biko benefit from the exchange? b. Did Ayanda benefit from the exchange? c. What is the rent that Ayanda receives as a result of this exchange? d. Did the exchange result in a Pareto improvement? e. Is the post exchange allocation (point h) Pareto efficient? 4.7 Bargaining power: Take-it-or-leave-it In the bargaining over the distribution of coffee and data above, the two examples of rules of the game were symmetrical. Neither "split the difference between the willingness to pay of the two" or "alternating coin flips to see whose preferred price will be used" gave any obvious advantage to either player But many bargaining interactions are asymmetrical. One of the players has most of the bargaining power. Bargaining power is the ability to gain a large share of the mutual gains from exchange (total rents) made possible from some interaction, as may be determined by the rules of the game governing the interaction and the skill of the players in securing a favorable agreement under these rules. An example is the Ultimatum Game in Chapter 2 (whose name already sug- B ARGAINING POWER is the ability to gain a large share of the mutual gains from exchange (total rents) made possible from some interaction, as may be determined by the rules of the game governing the interaction and the skill of the players in securing a favorable agreement under these rules. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 193 gests the asymmetry). The Proposer makes a offer of some fraction of the "pie". The Responder’s strategy set is simply: accept or reject, or "take it or leave it." Being in a position to make that kind of an ultimatum is called take it or leave it power, or TIOLI power for short. In the coffee for data bargaining game if Ayanda had TIOLI power, she could have said to Biko: “I’ll give you 2 kilograms of coffee and you give me 9 gigabytes of data. If you refuse, I will not agree to any other trade you might TAKE - IT- OR - LEAVE - IT- POWER A player with TIOLI power in a two-person bargaining game can specify the entire terms of the exchange – for example, both the quantity to be exchanged and the price – in an offer, to which the other player responds by accepting or rejecting. propose.” In other words, ”either accept the allocation I impose, or we both stay at our endowment, z." The assumption that Ayanda’s offer is credible is important: if Biko suspects that he could refuse and Ayanda would listen to a counter-offer, the threat in the TIOLI offer would be empty. A bargainer with TIOLI power can often capture most or even all of the total rents that an economic interaction provides. This is because TIOLI power allows a bargainer to specify both: • the price at which the goods will be exchanged and • the amount of goods that will be exchanged This means that the person with TIOLI power can just pick some preferred allocation – a point in the Edgeworth box different from the endowment point – and make that the TIOLI offer. What take-it-or-leave-it offer will Ayanda make to Biko? Ayanda does not care about Biko’s utility, but she does care about how he will respond to her offer. If he rejects, then she gets her fallback option. She will realize that she must offer Biko a deal that Biko regards as better – or at least not worse – than the endowment. In other words, Ayanda has to take Biko’s participation constraint as a limit on the kind of offer she will make. This is an example of the backward induction method that you learned in Chapter 2: Ayanda has to reason backwards from her understanding of what Biko will do after she has made her offer to what offer she should make now. So Ayanda has the following constrained maximization problem: find a final allocation (different from the endowments) to propose at which Biko is no worse off than at the endowment and Ayanda is as well off as she can be. Ayanda knows that the solution to this problem must have two characteristics: It must: • satisfy Biko’s participation constraint, that is, be in (or on the boundary of) the Pareto-improving lens in Figure 4.7. • be Pareto-efficient, but this is not because Ayanda cares any more about efficiency than she does about Biko: if she offered an allocation that satisfied Biko’s participation constraint and was not Pareto efficient then there R E M I N D E R The Ultimatum Game discussed in Chapter 2 has this TIOLI structure including returning to the endowment point the the Responder rejects – both getting a payoff of zero, namely what they would have received had they not interacted. That is why it is called the Ultimatum Game, as the Proposer’s offer is an ultimatum. 194 MICROECONOMICS - DRAFT would be some other allocation at which she could be better off and Biko not worse off. Ayanda would probably offer Biko something just a tiny bit better than Biko’s fallback utility to make sure he accepts. But to avoid having to keep track of that tiny amount in our thinking, here and in the rest of the book, we will assume that Biko will accept an allocation that just meets his participation constraint. That solves the problem for Ayanda: to meet the two requirements bulleted above , she must find the intersection of the Pareto-efficient curve and Biko’s participation constraint uB z . Therefore, Ayanda offers an exchange that imple- M - C H E C K Remember that in Chapter 3, a utility maximizer is often constrained by a feasible frontier. Even with TIOLI power, Ayanda is constrained by Biko’s participation constraint, that is, uB (xB , yB ) uB z. ments point tA at the extreme end of the Pareto-efficient curve in Figure 4.7 a. The same result is shown in Figure 4.7 b, where tA represents the distribution of utilities resulting from the TIOLI allocation that Ayanda offered and Biko (barely and grudgingly) accepted. Point tB at the other extreme of the Paretoefficient curve in the figure in the Edgeworth box corresponds to the allocation where Biko has TIOLI power and point tB on the utility possibilities frontier is the corresponding distribution of utilities. We can see that the TIOLI allocation does not weight the two utilities identically (as did the social welfare function of the Impartial Spectator, which led to point i). This why we say that allocation tA is Pareto-efficient but not socially efficient, where the latter term is whatever the Impartial Spectator selected based on maximizing an equally-weighted social welfare function. Two features of the TIOLI outcome where the participation constraint holds are important because they arise in many social coordination problems that involve a participation constraint: 1. Pareto efficiency: The PC-constrained outcome is Pareto-efficient. 2. Inequality : At PC-constrained outcomes the bargainer with TIOLI power gets all of the economic rent. M-Note 4.4: Finding Ayanda’s TIOLI Offer We need two pieces of information to find Ayanda’s TIOLI offer: • The equation for the Pareto efficient curve (because we know that the resulting allocation will be Pareto efficient) and • The equation for Biko’s participation constraint (because we know that Ayanda will not offer him anything better than his utility at his endowment bundle). The Pareto efficient curve: At Checkpoint 4.3 we asked you to find the Pareto-efficient curve for Ayanda and Biko when they have identical Cobb-Douglas utility functions with a = 0.5. The solution is that the Edgeworth box has the following Pareto-efficient curve R E M I N D E R An outcome is socially efficient when it maximizes a social welfare function; what is deemed socially efficient depends on how the utility of each member of the population is weighted in the social welfare function. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S defined over the two people’s allocations of x and y: yA ✓ ◆ 3 A x 2 = (4.3) (4.4) We can re-write Equation 4.5 in terms of xB and yB by substituting xA ȳ yB in the equation to find: yB = x̄ xB and yA ✓ ◆ 3 B x 2 = = (4.5) As you can see, the Pareto-efficient curve is a line from the one corner of the Edgeworth Box to the other. The utility functions, endowments and TIOLI offers calculated in this M-Note are the basis for Figure 4.7. To find the TIOLI offers, we need the players’ participation constraints because the player with TIOLI power wishes to maximize their utility subject to the participation constraint of the other player. Biko’s participation constraint: B’s fallback utility (his participation constraint (PC)) at his endowment xzB = 1, yB z = 14 is: uBz (1, 14) = (1)0.5 (14)0.5 = 3.74, So we need to find the point on the PEC at which Biko has this level of utility. A’s TIOLI Offer: We substitute the Pareto-efficient curve’s value for xA into B’s utility function that equal to B’s fallback utility: uB = (xB )0.5 ✓ ◆ 3 B 0.5 x 2 | {z } = uBz = 3.74 | {z } PC PEC 3 ) ( )0.5 xB 2 = B xTA = ) yBTA = A ) xTA A ) yTA = 3 3.74/( )0.5 = 3.05 ⇡ 3 2 3 B 3 9 x = (3) = = 4.5 2 2 2 x̄ xB = 7 = ȳ 3.74 yB = 10.5 So where “TA” means A had TIOLI power, the post-exchange allocation will be A = 7, yA = 11.5, xB = 3, yB = 4.5 . The post-exchange allocations imply that A xTA TA TA TA B made a TIOLI offer to B of 2 units of x (xTA y (yATA yAz = 14 xzB = 3 1 = 2) in exchange for 9.5 units of 4.5 = 9.5). A’s utility is uATA = 8.97 and B remains on his participation constraint at uB z = 3.74. Checkpoint 4.5: Biko’s TIOLI offer to Ayanda Given the TIOLI offer you just saw for Ayanda in M-Note 4.4, what would happen if the players’ positions were reversed and Biko had TIOLI power over Ayanda? To answer this here is A’s fallback utility at her endowment xzA = 9, yA z = 1: uAz (9, 1) a. What offer would Biko make? = (9)0.5 (1)0.5 = 3, 195 196 MICROECONOMICS - DRAFT b. What would the post-exchange allocations be? Explain. c. Use Ayanda’s utility at her fallback position that we found in M-Note 4.4. 4.8 Application: Bargaining over wages and hours We illustrate TIOLI power below by a case in which the two bargainers drop their student personas to take on familiar roles in what is arguably the most important market in a modern economy: Ayanda is the owner of a company interacting with Biko, a prospective employee. In labor market bargaining over wages and working conditions the employer almost always has TIOLI power, stating the wage, the job and the hours. The worker accepts or not. We postpone until Chapter 15 the question: why might Ayanda get to have this power and not Biko? So, leaving the world of coffee and data behind us, we will see that the sum of the mutual gains enjoyed by the two and how these are divided between them will depend on both their preferences and the rules of the game: • Power: Do the two bargain symmetrically with neither one nor the other of them having first mover advantage? Is one of them first mover with take-itor-leave-it power (TIOLI power)? • Fallback: What is each person’s fallback position? how well off are they if they do not exchange at all? Does Biko have other options than being employed by Ayanda? If Ayanda does not employ Biko, are there others she she could employ? To fill in some answers to those questions, our two actors are now: • Ayanda, an employer : whose endowment bundle is a sum of money only (no employees), and who in the absence of any exchange with Biko has nobody work for her; she will make Biko a take it or leave it offer of a sum of money in return for some number of hours of work for her, and • Biko, a worker : who is applying to work in Ayanda’s company, whose endowment bundle is free time only (no money); he has a maximum of 16 hours of (non-sleeping) time to spend, possibly working for Ayanda. We introduce a more complete model of the labor market with competition among firms for workers and customers and among workers for jobs in Chapter 11 including the ways that unemployment benefits, competition among firms, and antitrust could affect these outcomes. Quasi-linear preferences for money and time To represent the preferences of Ayanda and Biko we will introduce a new form of utility function, one that will simplify our analysis while still conveying Q UASI - LINEAR FUNCTION A quasi-linear function depends linearly on one variable, e.g. y, and non-linearly on another variable, e.g. x, and has the form u(x, y) = y + h(x), where h(x) is a non linear function. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S Figure 4.8: Marginal rates of substitution with quasi-linear preferences. With quasi-linear preferences, marginal rates of substitution depend only on the amount of the good x (here, Hours of Living for Biko), and not at all on the amount of money left over to buy other goods, y. As a result, indifference curves with different levels of utility are vertical displacements of a single curve – you can add or subtract an amount of y from the indifference curve to move it up or down. y3 = 400 y2 = 340 Quantity of money, y 197 y1 = 260 h g e uB3 uB2 f uB1 0 0 2 4 6 8 10 Hours of Living, x 12 14 16 the main insights. The function is called quasi-linear because utility is partly (“quasi") proportional to one of the arguments of the function, while being nonlinear in the other arguments. The Cobb-Douglas utility function is not quasi linear because it is non linear with respect to both x and y. As in the case of Harriet deciding how much fish to buy from Alfredo or Bob in Chapter 3, we will consider the second good as “money left over" after the exchange. This may seem odd because money is not something you value for itself. But money can buy you other goods which you do value: The utility of "money left over" is the utility of the goods which the person can purchase as a result. We now illustrate a case where one person starts off with all of one good and none of the second, so the other person all of the second good,but none of the first. This could model you walking into the supermarket with money in your pocket (or more likely a credit card) and nothing in your shopping bags, and planning to walk out with less in your credit card and some groceries in our shopping bag. So it is a model of any kind of exchange. But here illustrate it by Ayanda (possibly) employing Biko. The marginal rate of substitution for a person with quasi-linear preferences that are linear in "money" (y) depends only on the amount she has of the good or service for which her preferences are non-linear, not on the amount of money. The reason why this is true is because: 198 MICROECONOMICS - DRAFT • The marginal rate of substitution is the ratio of the marginal utility of x to the marginal utility of y • The person’s marginal utility for y is always a constant and it does not decline as she gets more y, so • The marginal rate of substitution depends only on the marginal utility of x which varies with the quantity of x consumed because the function is non linear in this variable. You can see this in Figure 4.8 by noticing that for a given amount of x the slope of the indifference curve (shown by the dashed tangent lines) is the same no matter how much y the person has, such as at x = 8 hours of living, as shown by points f, g, and h. This is because, given the quasi-linear utility function that we used to draw the figure, the willingness to pay for an additional hour of living (the marginal rate of substitution, that is, the negative of the slope of the indifference curve) does not depend on the amount of money lef over that the person has; it depends only on how many hours of living they have. This means that the indifference curves u1 , u2 , and u3 in the figure are just shifted up replicas (you can see the amounts by which they are shifted up by comparing the vertical axis intercepts). We can also compare points e and g at the same level of y: Biko likes to have B more living time (uB 1 < u2 ) and his willingness to pay for additional hours of Living declines the more he has (the indifference curve is less steep at g than at e). It is of course unrealistic to think that anyone would have truly linear preferences in any amount imaginable of money left over, for this would require that the person did not have diminishing marginal utility in the things that money can buy. But because "money" can be considered as generalized purchasing power that can be spent on a vast array of things, and because we do not consider changes in people’s bundles of money making them either billionaires or paupers, it is a useful simplifying assumption. M-Note 4.5: A quasi-linear utility function Quasi-linear utility functions have the form: u(x, y) = y + h(x) (4.6) Equation 7.29 is quasi-linear because it is linear in one variable, y, and non-linear in the other x as in the non linear function h(x). For example, h(x) could be quadratic in x or could include the natural log of x, ln(x). A quasi-linear utility function depends linearly on one variable, e.g. y, and non-linearly on another variable, e.g. x, and has the form u(x, y) = ay + h(x), where a is a constant. Hence it is quasi or ‘partly’ linear. We often set a = 1. Two examples of such functions include, u(x, y) = y + 20x x2 and u(x, y) = y + 10 ln(1 + x). P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 199 Allocating money and time For simplicity, we assume that Ayanda, the employer, and Biko, the worker, have quasi-linear utility functions. Both place a constant value per monetary unit on the money they have after the transaction (to purchase other things, for example). That is, the marginal utility of money is constant. So we can measure the utility of each in whatever monetary units they are using, which since their names are from South Africa, might as well be the South African Rand. Biko values his Living (that is his 16 waking hours, minus the time he "hires out" of himself to work for Ayanda). But the marginal utility of free time is decreases as the amount of free time he has increases, just as was the case for Aisha in Chapter 3. His first hour of free time is high, but his free time gets less and less valuable the more free time he has. Ayanda places a value, too, on Biko’s free time, but it is the opposite of Biko’s value: she benefits by Biko having less free time and her having more of Biko’s time working for her. The positive value she places on Biko’s labor – like the positive value he places on his free time – depends on how much of it she gets. The marginal utility to Ayanda of Biko’s labor decreases as she hires more of his time: the value of Biko’s work is high the first hour Ayanda hires, less valuable the second hour, less valuable the third, and so on. This is because if she has just an hour of his time, she assigns him to really important tasks, but the tasks he does in later hours are less essential to Ayanda. (This is similar to why the marginal productivity of time studying diminishes as the amount of time studying increases). Not accounting for the wage she must pay him, each hour of Biko’s work gives Ayanda utility, but at a decreasing rate. Figure 4.9 shows the setting for this interaction as an Edgeworth box, with the quantities interpreted amounts per day. The endowment point z is in the upper left corner of the box showing that Biko has 16 hours of Living time and no money. Ayanda has $400 but no Labor from Biko to work in her company. As before, like z every point in the box represents an allocation that is feasible given the amount of money that Ayanda has in her endowment bundle and the amount of free time that Biko has in his. Three of Biko’s indifference curves and three of Ayanda’s are shown in Figure 4.9. For both Ayanda and Biko, their reservation indifference curve (their participation constraint) goes through the endowment point where Biko has 16 hours of Living and Ayanda has $400 per day to pay workers, meaning that the indifference curves include the endowment point z. Also shown is one of Biko’s indifference curves labeled uB 3 , which is tangent to B Ayanda’s participation constraint (uA z ) at point t . The allocation given by the F AC T C H E C K At the time of writing this 1 Euro was equal in value to about 16 South African Rand (ZAR), 1 Pound Sterling was equal to about 21 South African Rand. In 2020, the hourly minimum wage in South Africa was ZAR 20.76. MICROECONOMICS 14 12 B's Hours of Living,xB 10 8 6 4 2 Biko 0 0 16 - DRAFT z uBz = 256 B's PC tA uB3 = 508 100 uA2 = 516 tB 300 200 j 200 Pareto−efficient curve uB2 = 392 A's PC uAz = 400 0 100 B's Money, yB A's Money, yA 300 uA3 = 652 0 Ayanda 2 4 6 8 10 12 A's Hours hired of B's Work, xA Figure 4.9: Bargaining over hours and wages. Shown are three each of Ayanda’s and Biko’s indifference curves and the utility that they experience at any of the allocations indicated by the points making up these curves. Point z is the endowment allocation which is a point on the participation constraints of each of the two. Points tA and tB respectively are the allocations resulting when Ayanda or Biko are first mover with TIOLI power. The yellow shaded area is the Pareto-improving lens. The vertical line (including its dashed portions) is the Pareto-efficient curve made up of all points of tangency between the indifference curves of the two such as j, tA and tB . 14 16 400 400 200 that tangency is a Pareto-efficient allocation (because the marginal rates of substitution of the two are equal). We also show a third indifference curve for B Ayanda, labeled uA 3 , which is tangent to Biko’s participation constraint (uz ) at point tA . These two tangencies are points on the Pareto-efficient curve, which is a vertical line through these points all the potential tangencies above each person’s fallback. The reason why the Pareto-efficient curve is vertical here (remember it was a diagonal line in the previous Edgeworth boxes) is that Ayanda and Biko have quasi-linear utility functions. With quasi-linear utility, the marginal utility of hours depends only on the quantity of hours and not on the amount of money they have. If the two curves are tangent at 8 hours when Ayanda has most of the money and Biko little, they will also be tangent at 8 hours when Biko has most of the money and Ayanda has little. 4.9 Application. The rules of the game determine hours and wages The Edgeworth box and the indifference curves by themselves do not determine the outcome of the interaction. Without knowing more, any point in the box is a possible outcome. Knowing the endowment allocation z narrows down the possible post exchange allocations but not by very much. Employment in most modern economies is voluntary (but see the Fact Check), so we will assume that the outcomes are limited to those that are at least as good for each participant as their fallback position given by point z. As a result, outcomes of bargaining between the employer and the worker must be in the yellow shaded Pareto-improving lens in Figure 4.9. F AC T C H E C K In the past slavery has meant the ownership of one person by another, including the right of sale of the slave to another owner. The term modern slavery refers to any situation in which, like historical slavery, the services or goods that one party provides for another are not voluntarily offered but are motivated by fear of severe harm. Ownership of one person by another need not be part of modern slavery. Prisoners, immigrants without legal rights of residence, residents of undemocratic countries, “sex slaves," and children are over represented among contemporary ’modern slaves.’ P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 201 We illustrate the importance of institutions by showing the allocations will result under four different rules of the game. Each set of rules is a specific account four different ways that an employer and worker might interact. : • The employer can make a take-it-or-leave-it offer of both the wage and the hours worked. • As members of a trade union, the employees (we will take Biko as a representative worker) can make a take it or leave it offer specifying both the wage rate and the length of the working day (hours). • Legislation is passed limiting working hours per day to no more than 5 hours and the total pay or this period to not less than 254 Rand or 50.80 Rand per hour. • The above legislation is passed, but it has a proviso that if the two parties can agree on an alternative allocation their agreement can be implemented. Employer has TIOLI power Imagine that, like most employers, Ayanda can offer Biko a job description: work a given amount of hours for a given amount of pay (and therefore for a particular hourly wage). Biko’s only choice is to accept or reject, so Ayanda has take-it-or-leave-it power. For Biko to accept, Ayanda knows the offer must be at least as good as Biko’s reservation option, so the relevant constraint for her is Biko’s participation constraint (as was the case for the coffee and data bargaining). She will choose the point that she values most along this indifference curve, and therefore implement an offer indicated by point tA . Having TIOLI power, the employer has gotten all of the economic rent, leaving Biko indifferent between taking the job and refusing it (as before in cases like this we just assume he takes the job). What is Ayanda’s rent from this transaction, meaning the excess of her utility at point tA compared to at point z, the endowment allocation at which no trade has occurred? Reading the utility numbers from her indifference curve at point tA and her reservation indifference curve through point z we can see that her A rent is uA 3 = 652 minus uz = 400 or 252. Because utility is measured vertically in terms of money this is the same thing as the vertical distance between points tA and tB in the graph. Employees and their trade union have TIOLI power Turning to the opposite case Biko, through his trade union, is now first-mover with TIOLI power. The offer he will make (and she will accept) is the opposite E X A M P L E Put yourself in Biko’s shoes if the allocation is point tA . How do you think he feels about his employer and his job? Would he be motivated to work hard, not to steal from his employer, and otherwise contribute to the profitable operation of her firm? These are serious problems and a reason why extreme allocations – like Ayanda getting all of the economic rent from the interaction and Biko being indifferent between his job and being fired – are not commonly observed. If Ayanda has an interest in Biko’s good will and hard work, she may have to share at least a bit of the gains from exchange with Biko so that he receives a rent. This fact will become important when we consider the labor market. MICROECONOMICS 12 B's Hours of Living,xB 8 Biko 4 0 0 16 - DRAFT z 100 uA3 = 652 Pareto− efficient curve a 200 A's Money, yA b B's Money, yB uB2 = 351 tA 200 300 uBz = 256 100 uB3 = 508 300 uA2 = 540 tB decreased work hours 0 uAz = 400 0 Ayanda 4 8 12 A's Hours hired of B's Work, xA 16 400 400 202 of tA the allocation resulting when Ayanda had TIOLI power. Biko will recognize Ayanda’s participation constraint – he as to make her an offer she will not refuse. And he will choose the allocation indicated by point tB in which his post exchange bundle gives him all of the economic rents of 252. This is the most that Biko could demand without Ayanda simply going out of business. This constraint on the demands that workers can make on employers in a market and profit based economy will be a major theme in the chapters to come. Legislation imposes hours and pay limitations. As described above the legislation imposes on both Ayanda and Biko the allocation at point b. But the allocation imposed by the legislation is Pareto inefficient. But it does set a new status quo, a fallback position that, if they cannot come to some agreement will be the post exchange allocation. Both Ayanda and Biko can see that at b they could both do better by agreeing that Biko should work more than 5 hours, and Ayanda should pay him more. The yellow Pareto-improving lens shows the space for their possible bargains. Bargaining to override the legislation: more work and more pay. They could bargain to agree upon any point in the yellow Pareto improving lens, possibly agreeing on the Pareto efficient allocation at point a. Where they ended up in or on the boundary of the Pareto improving lens would depend on the rules of the game governing that bargaining process. They might even fail to agree on any bargain – as is often the case with players in Figure 4.10: Allocations with legislation and bargaining. The legislation stipulating hours and pay results in the allocation indicated by point b. Because b is preferred to the no exchange option z by both of them, they will definitely make an exchange. But then can both do better than at b. Taking the allocation at b as their new fallback position, they could bargain to point a or any other allocation in the yellow Pareto improving lens. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S the Ultimatum Game – and remain at point b. Figure 4.11 shows how Ayanda and Biko do under these differing rules of the game as indicated by the rents they enjoy in the Nash equilibrium of each game, that is the utility associated with their fallback options 400 and 256 respectively. Introducing a historically realistic set of rules of the game – making the employer the first mover with TIOLI power – has two effects: it generates 252 units of utility in gains from trade, and it makes the final allocation more unequal than the endowment allocation (because the employer captures all of the mutual gains made possible by exchange). Biko’s share of the total utility (not shown) falls from two fifths to less than a third. In many countries during the 20th century the response to the unequal allocations implemented when the employer has TIOLI power was the formation of trade unions. And you can see from the figure that if the union is powerful enough for it to have TIIOLI power, then Biko (and his trade union colleagues) capture the entire rent, Ayanda getting nothing more than her reservation utility. Biko’s share of the total utility jumps from two fifths at the endowment allocation to well over half. Even before workers had the right to vote and before trade unions were legal, political movements mobilized to pressure governments to regulate working conditions. In the model the introduction of hours and wage regulations implemented an outcome in which both Biko and Ayanda captured some of the gains from trade. The reforms implemented a Pareto inefficient allocation but the shortfall from the maximum possible joint rents was minor (from 252 to 235). The final case – bargaining up from the regulated hours and wages – describes labor markets in many countries today. Government regulations establish a fallback position, and then employers and workers (either individually or in trade unions) seek bargains that improve on that allocation. Though they differ radically in their distributional aspect, all of the scenarios are Pareto superior to the endowment allocation. We can also see that the negotiated allocation after legislation is Pareto superior to the allocation implemented by the legislation. We cannot say which of the three Pareto efficient allocations is preferred from a fairness standpoint without knowing more about Ayanda and Biko’s other wealth, their needs, and other aspects that might affect their ethical claims on the benefits of their interaction. Checkpoint 4.6: Bargaining over hours and wages 203 204 MICROECONOMICS - DRAFT Type B's rents A's rents Gains from exchange 0 Employer (A) has TIOLI power, t A Union (B) has TIOLI power, t B 252 252 252 0 Figure 4.11: Rents under differing rules of the game, with Ayanda as employer and Biko as worker The rents and gains from exchange of each set of rules are shown in the figure. That is, the figure shows each player’s utility under each set of rules minus that player’s fallback option (uAz = 400 and uBz = 256 respectively). The gains from exchange are the sum of the rents received by Ayanda and Biko. Source: Authors calculations described in the text. 252 95 Legislated hours and wages, b 140 235 104 Negotiated allocation after legislation, a 148 252 0 100 Rents 200 300 Using the figure, explain how the following two things (taken separately) would affect the outcome under the four different rules of the game above (start by explaining how the endowment point z would be affected): • If Biko does not exchange their time with Ayanda and is unemployed, he receives what is called an unemployment benefit, that is a payment from the government equal to $100, and this is financed by a tax on Ayanda equal to $100. • Ayanda now has free access to a robot that will at no cost do work equivalent to two hours of Biko’s time. 4.10 First-mover advantage: Price-setting power Returning to Ayanda and Biko with their former personas as students exchanging coffee and data, we will now see that while first movers typically have advantages, these advantages may not be due to TIOLI power. Ayanda may be first-mover but be unable to commit to take-it-or-leave-it offer that stipulates an exchange of a specific amount of coffee for a specific amount of data. Price-setting power She may have what is called price-setting power (PS power) if she can specify a price – either a monetary price or the ratio at which the two will exchange goods – but not how much (the quantity) of her good Biko will buy. Ayanda might say, for example: “I will give you one kilogram of coffee for every three gigabytes of data you give me. You can decide how much data F IRST- MOVER ADVANTAGE A player has a first-mover advantage when the institutions, history, or power structures of a game give the player the opportunity to make an offer or move before the other players in the game can take action. The opportunity to move first can confer an advantage that results in higher utility or a greater share of economic rents in the outcome of an interaction. P RICE - SETTING POWER A first-mover with price-setting power (PS power), can commit to a price – the ratio in which goods will be exchanged – but not the quantity that will be transacted at that price. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 205 you would like to exchange for coffee at that ratio, but the ratio is not going to change. Of course you are free to buy nothing.” We saw that owners of companies typically have TIOLI power when hiring employees; but in their interactions with their customers they typically have price-setting power. They set a price at which they will sell their product, and sell as much to each customer at that price as the customer wants to buy. The incentive compatibility constraint (ICC). If Ayanda has price setting power she must find a way to determine the price when it is the price alone that makes up her offer. So her constrained optimization problem is not the same as it was when she had TIOLI power. When Ayanda had TIOLI power she had only to satisfy Biko’s participation constraint: if her take-it-or-leave-it offer were a post exchange bundle that would make Biko worse off than at his endowment bundle, Biko would leave it rather than taking it! Of course whether she has TIOLI power or just price setting power, if Ayanda wants to exchange with Biko, she will have to satisfy his participation constraint. But there is now a second constraint she must satisfy called the incentive compatibility constraint. What this means is whatever post-exchange bundle Ayanda would like to implement, she must provide Biko with incentives so that his best response will be to exchange the amount that will allow her to "move" from her endowment bundle to her desired post-exchange bundle. This is called the incentive compatibility constraint because she must provide Biko with incentives that motivate Biko to act in a way that is compatible with (meaning, that implements) her desired outcome. The incentive compatibility constraint is based on Biko’s best response – the amount of coffee he is willing to buy – to the price Ayanda offers. You have encountered best responses in Chapter 1. There the the strategy sets were particular actions and therefore best responses were limited to actions like “Plant Late," or "Fish 12 hours." Options like "Plant a little earlier" or "Fish 10 hours and 15 minutes" were not possible. Sometimes discrete strategy sets and best responses like this make sense (think: "Drive on the the left if you are in the U.K., or Japan"). But sometimes the strategy sets for players are continuous, as for example in setting a price for a good or when choosing the amount of time for an activity, like fishing. When this is the case – as with Ayanda’s decision to set a price – we consider the players’ best responses as continuous variables and describe them by best-response functions. M - C H E C K A continuous variable can take on any value over some interval. So, a variable that can take the value of any number between 0 and 5 is a continuous variable; a variable that is restricted to the integers between 0 and 5, namely, 1, 2, 3 or 4 is discrete. The number of your sisters or brothers is discrete, the height any one of them is continuous. I NCENTIVE C OMPATIBILITY C ONSTRAINT The incentive compatibility constraint, ICC, requires that first mover provide incentives that make the second mover’s best response be to act in ways that implement the post exchange allocation which the first mover prefers. 206 MICROECONOMICS - DRAFT As was the case when she had TIOLI power, Ayanda will reason backwards from her understanding of what Biko how Biko will respond to each of her possible offers and how that will affect her utility. That is, she will use backward induction. To determine how Ayanda can maximize her utility subject to Biko’s incentive compatibility constraint (the price-setting case) is a somewhat more complex problem than maximizing her utility subject only to Biko’s participation constraint (the TIOLI power case). The reason is that in the TIOLI case there are just two things that Biko can do: accept or reject her offer. But when Ayanda has price-setting power only, Biko can choose from the entire range of possible amounts that he might be willing to exchange with her, depending on the price. As a result, Ayanda has to think in two stages when choosing a price ratio. First stage: What will Biko do? How much coffee will Biko buy at each price ratio Ayanda offers? This is Biko’s price-offer curve, which is his best response. Second stage: What should I do, given what he will do? Given her estimate of Biko’s best response, which price ratio maximizes Ayanda’s utility? That is, which price ratio takes Ayanda to her highest indifference curve, given the constraint of Biko’s price-offer curve? Best response and incentive compatibility For the first stage, that is, determining how Biko will respond to each price she might offer, Ayanda uses whatever information she might have, such as her experience in the past with Biko’s response to offers, her best guess as to Biko’s utility function, or her experience with other people she thinks are similar to Biko. Just as in Chapter 3 there is a budget constraint limiting the exchanges he can undertake, but this is now a line giving feasible combinations of data and coffee available to him through exchange at some given price. If the price p – the number of gigabytes of data per kilogram of coffee – and his post-exchange bundle is denoted as (xB , yB ) then Biko’s budget constraint requires that the value of his post-exchange bundle must be the same as the value of his endowment bundle, or: pxB + yB = px̄B + ȳB or p(x B B x̄ ) = B (y B ȳ ) (4.7) (4.8) The second version of the budget constraint means that the value of the coffee that he acquires (at the price p) or xB x̄B must be equal to the value R E M I N D E R The method is identical to how we derived Keiko’s price-offer curve – offering money in return for fish – in Chapter 3, except that here Biko is not ’buying’ coffee using money, he is exchanging data for coffee. As a result the "price" is not in terms of dollars per kilogram of coffee, but gigabytes of data per kilogram of coffee. P R O P E RT Y, P OW E R , of the data that he gives up ȳB & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S yB . We can rearrange Biko’s budget constraint another way to show that the price p must be equal to the ratio of the amount of data he gives up to the amount of coffee he gets p= ȳB xB yB x̄B (4.9) We show the derivation of Biko’s best-response function in Figure 4.12. We start, in panel a by showing Biko’s best response to one particular price. Then, in panel b we repeat the same reasoning for many prices, showing how his best response to any price can be determined. We know that given the price p4 Biko will choose how much data to transfer to Ayanda in return for her coffee in order to maximize his utility subject to his budget constraint. In panel a we show his feasible set with his budget constraint for that particular price p4 its feasible frontier. The budget constraint includes the point z because one of the feasible choices he could make while respecting the budget constraint is to exchange nothing. In Figure 4.12 the slope of the p4 line is the amount of data that Biko gives up (DyB ) divided by the amount of coffee that he gets (DxB ) , when the price is p. So: p = DyB ȳB = B B Dx x yB x̄B marginal rate of transformation (mrt) = slope of the price line For any given price this is the kind of individual utility maximization problem that you studied in Chapter 3 in which the solution is to find the allocation at which the mrs = mrt rule holds. You can see in panel a that the highest indifference curve that Biko can reach, consistent with his budget constraint (labeled uB 2 ) is tangent to his budget constraint at point b4 . This result expresses the principle of constrained optimization that you have already learned. It is a point equating: • The slope of his indifference curve, which is the negative of the marginal rate of substitution and • The slope of the feasibility frontier – in this case the budget constraint – which is the negative of the marginal rate of transformation of coffee into data. The mrt is the price p set by Ayanda, that tells Biko how many gb of data he has to give up to get a kilo of coffee. Biko’s best response is to choose a 207 MICROECONOMICS 8 B's coffee (kilograms), xB 7 6 uBz , B's PC 5 4 3 2 1 Biko 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 p4 B's feasible set uB2 Coffee B gets b4 Data B gives up z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 10 10 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 8 p2 b2 4 3 2 1 Biko 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 uB3 b3 b4 uB4 B's best− response function (ICC) z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 10 (b) B’s best-response function (ICC) (4.10) and, budget constraint: pxB + yB = px̄B + ȳB abbud (4.11) Equation 4.10 expresses the optimizing part of Biko’s choice, while Equation ?? expresses the constraint. The utility Biko enjoys at b4 in the figure is B greater than the utility of his endowment bundle (uB 2 > uz ). From this we conclude that if the price is p4 Biko will choose the post-exchange bundle given by point b4 . This gives us one point on Biko’s best-response function. In panel b we construct Biko’s best response function, by repeating the analysis in panel a but for differing prices tracing out a curve in the (x, y) coordinates. This is his best-response function because, by construction, points on the curve show for each the value of p the post-exchange allocation that maximizes his utility if could buy any amount of Ayanda’s coffee at the price p. Ayanda now has all the information she needs to set the price. M-Note 4.6: The incentive compatibility constraint Here we show the derivation of the incentive compatibility constraint for Ayanda’s utilitiy choice of a utility maximizing price to offer Biko. This equation will show, for every price that Ayanda could offer, the amount of goods that Biko will be willing to exchange. To do this we use Equations ?? and 4.10, the two conditions that Bikos response must satisfy. Given the price p offered by Ayanda, Biko’s budget constraint is yB ( x B ) = 5 uB2 mrs = mrt tangency: mrsB (xB , yB ) = mrt = p (4.12) pxB + px̄B + ȳB To maximize his utility uB (xB , yB (xB )), Biko will choose the bundle 6 p4 post-exchange bundle that satisfies the two conditions: That is 7 p3 (a) B’s best response to a price = p4 pxB + yB = px̄B + ȳB B's coffee (kilograms), xB B's data (gigabytes), yB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 B's data (gigabytes), yB A's data (gigabytes), yA 10 - DRAFT A's data (gigabytes), yA 208 (xB , yB (xB )) satisfies Figure 4.12: Constructing B’s best-response function (ICC). In panel a, B’s feasible set is in the upper right corner of the Edgeworth box because, as we explained in Figure 4.4, the upper left corner of the box is the origin for him (indicating zero of both goods). In panel a, when the price p4 is equal to 3.53 Biko reaches his highest feasible indifference curve (uB2 ) by giving up 5.3 gb of data in return for 1.5 kg of coffee. In panel b he chooses post-exchange bundles indicated by points b3 and b2 in response to prices p3 < p4 and p2 < p3 . B’s best-response function (ICC) connects these and similar points all of them B’s utility-maximizing bundle, for different prices. P R O P E RT Y, P OW E R , the first-order condition, uBx + uBy dyB = uBx dxB That is 1 uBy p = 0 uBx = uBy mrsB (xB , yB ) = & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S p = mrt (4.13) 2 Suppose that uB = (xB ) 3 (yB ) 3 , we can derive the incentive compatibility constraint using Equations 4.12 and 4.13. From M-Note 4.2, we have mrsB (xB , yB ) = 1 yB 2 xB Moreover, the budget constraint can be rewritten as Equation 4.9, i.e., p= Therefore, we have 1 yB = 2 xB ȳB xB yB x̄B ȳB xB yB x̄B (4.14) which defines the incentive compatibility constraint shown in the Edgeworth box. 4.11 Setting the price subject to an incentive compatibility constraint Biko’s best-response function is the incentive compatibility constraint for Ayanda’s optimizing problem, shown in Figure 4.13. Notice that the incentive compatibility constraint is more limiting to Ayanda than is Biko’s participation 0 constraint labeled uB z , B sPC. This means that there are some allocations (between the participation constraint and the incentive compatibility constraint) which would make Biko better off than at his endowment bundle, and which Ayanda would prefer to any point in her feasible set, but which Ayanda could not implement when she has price-setting power but not take-it-or-leave-it power. Because Ayanda always has the option of simply discarding some of the data she gets from Biko, we can think about the green shaded area under Biko’s best response function as her feasible set. The slope of Biko’s best-response function is (from Ayanda’s viewpoint) the marginal rate of transformation of coffee into data, given how Biko responds to each of the prices she could set. You can see that starting at the endowment allocation, the best-response function is initially steep, so a modest amount of coffee that she gives up is transformed – through exchange – into a substantial amount of data. But the more data she wishes to acquire – moving up on the best-response function – the less favorable to her the mrt becomes. Ayanda’s choice of what price to set is a familiar constrained optimization problem. It proceeds in two steps: 1. Determine the final allocation she would like to implement by finding the point in the feasible set that is associated with the higher utility. To do this she uses the mrs = mrt rule and selects point n in the figure, with its associated utility uA N (which exceeds that associated with point w, namely 209 MICROECONOMICS 8 B's coffee (kilograms), xB 7 6 5 4 3 2 Biko 1 0 uBz , B's PC w n uAN B's best− response function (ICC) A's feasible set uAw uAz z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 (a) Maximizing utility subject to the incentive compatibility constraint 10 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 B's coffee (kilograms), xB 8 7 6 5 uBz , B's PC 4 3 2 Biko 1 0 pN uBN tA Pareto− improving lens n uAN B's best− response function (ICC) uAz z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 B's data (gigabytes), yB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 9 B's data (gigabytes), yB A's data (gigabytes), yA 10 - DRAFT A's data (gigabytes), yA 210 10 (b) Pareto-superior alternatives to Ayanda’s choice of n uAw , which was also feasible. This is where her indifference curve is tangent to Biko’s best best response function. This is shown in Panel a of Figure 4.13 2. Determine the price that will implement this outcome. Every allocation on the best response function corresponds to some particular price that will implement it. Price pN shown in Panel b of Figure 4.13 implements point n We have given the price that Ayanda sets a superscript N because the allocation that it implements is a Nash equilibrium. To confirm that this is the case we ask two questions: • Given the strategy that Ayanda has adopted – that is, setting the price pN – is there any way that Biko do better than he does by trading with her so as to implement her chosen allocation (point n)? The answer is no, because n is a point on his best response function, which tells us that if she offers the price pN the best he can do is to trade with her so as to implement her desired point. • Given the strategy that Biko has adopted – his best-response function – is there any way that Ayanda could do better than she does by setting the price pN ? The answer is no, because she found point n exactly by doing the best she could given his best-response function. There are two important aspects of the Nash equilibrium (n) of this game. First, the Nash equilibrium is not Pareto efficient. Ayanda’s and Biko’s indifference curves are not tangent at n, they intersect, and you know from the mrsA = mrsB rule for a Pareto efficient outcome that any allocation at which the indifference curves intersect is not Pareto efficient (because then the rule is violated). The reason why Ayanda implemented an Pareto inefficient alloca- Figure 4.13: A sets the price subject to B’s best-response function (ICC) Ayanda’s utilitymaximizing post-exchange bundle is indicated by point n where her indifference curve is tangent to Biko’s best-response function (his price-offer curve). The negative of the slope of the solid gray line through both n and the endowment point z is equal to the price Ayanda chooses, pN . Biko’s budget constraint given by Ayanda’s choice of pN is tangent to Biko’s indifference curve through n by construction, that is, because n is on Biko’s best-response function. To interpret the lower shaded area as a feasible set, it must be the case that A could choose not to consume the data or coffee she has in that area (that is, some of it could be thrown away). P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S tion is that the the constraint she faced was not Biko’s PC (the slope of which is mrsB ) but instead his best response function (the slope of which is the mrt ). So she implemented mrsA = mrt 6= mrsB violating the Pareto efficiency rule. The allocations that are Pareto-superior to n are shown by the yellow lens between the indifference curves through n. Second, the person who is not the first mover (Biko) receives a rent in the Nash equilibrium: as you can see from Figure 4.13 at n he is better off (on a higher indifference curved) than with his endowment bundle (which is his fallback option, namely no trade) indicated by the indifference curve labeled uBz , B’s PC. These two results of the first mover with price-setting power only case contrast with the case of the TIOLI power. There is an important lesson here: when one of the two parties has price-setting power, but not TIOLI power, she may use that advantage to advance her distributional interests in a way that implements an inefficient outcome. For example, Ayanda could have implemented a Pareto-efficient outcome, like point w shown in the figure. This point is on the Pareto-efficient curve (not shown in the figure), and had she offered Biko the lower price given by the slope of a budget constraint from point z to point w, he would have purchased just the amount of coffee that would have implemented point w. But her utility is higher at point n which she can implement by charging a higher price (steeper budget constraint for Biko). This explains why it is the case that When Ayanda has price-setting power only she uses it to get a larger piece of a smaller pie. When she had TIOLI power she knew that she would get the maximum economic rent (because the only constraint she faced was Biko’s participation constraint). Subject only to the participation constraint she could dictate the entire outcome, so she had no reason to adopt any allocation that was not Pareto efficient. We will see that this feature of the price-setting case also reappears in other economic interactions – including credit markets, labor markets, and markets for goods with limited competition. Checkpoint 4.7: PSP vs. TIOLI a. Using Figure 4.13, by reading the relevant points on the x and y axis, say what the post-exchange allocations for Ayanda and Biko (how much coffee for each, how much data for each). Compare this to the post-exchange allocations when Ayanda has TIOLI power, calculated in M-Note 4.4. b. Test your understanding of the first-mover case by explaining the outcome when Biko is the first-mover. Draw a new version of Figure 4.13. 211 212 MICROECONOMICS 4.12 - DRAFT Application. Other-regarding preferences: Allocations among friends Ayanda and Biko are about to experience one final change in their identities, along with a personality transplant: they have become friends and they care about each other. Both are altruistic: they place some positive weight on the well being of the other. This means, as you will recall from Chapter 2 that they are other regarding, when evaluating an allocation they take account not only of the utility they will experience from their bundle but also the utility the other will experience from their bundle. They still have a decision to make: how to divide up their coffee (still 10 kilos of it) and the data (15 gigabytes of it as before). But we will assume now that neither of them own any portion of either good – so there is no endowment allocation like our interpretation of point z so far. The see how the Edgeworth box helps us to understand their decision problem and because this involves some unusual indifference curves, we first treat a hypothetical case in which Ayanda is completely altruistic and Biko is as before entirely self-regarding. (We do not imagine that Ayanda would put up with this, it is just a first step along the way to seeing how two other regarding friends would look at the problem). An altruistic utility function Altruistic Ayanda cares not only about her bundle at an allocation, but also what Biko gets. Ayanda’s utility therefore depends not only on xA and yA but also on xB and yB . We measure how much she cares about what Biko gets – her degree of altruism – by l ( "lambda") a number that varies from 0, if she is entirely self regarding, to one-half if she places as much weight on what Biko gets as what she herself gets, in which case she would be called a perfect altruist. M-Note 4.7: An altruistic utility function Remember if Biko did not exist so that Ayanda were making her choice of an allocation in isolation, her utility would be uA (xA , yA ) = (1 a ) xAa yA But interacting with Biko and dividing goods with him, for l function as an altruist: uA (xA , yA , xB , yB ) = ⇣ (1 xAa yA (4.15) > 0 we have Ayanda’s utility ⌘ ⇣ ⌘ a ) (1 l ) (1 a ) l xBa yB (4.16) To see why we say that l is a measure of how much Ayanda cares about what Biko gets we can take the natural logarithm of equation 4.16 P R O P E RT Y, P OW E R , ln(uA ) = (1 ⇣ (1 l )ln xAa yA a) ⌘ ⇣ (1 + l ln xBa yB a) ⌘ & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S (4.17) Equation 4.17 says that the natural logarithm of A’s utility is (1 l ) times the natural logarithm of her valuation (if made in isolation) of her own bundle plus l times the natural logarithm of B’s evaluation (if made in isolation) of his bundle. Biko’s utility function a has the same structure as Ayanda’s but the interpretation of l is the opposite. In Biko’s utility function l is the exponent of Ayanda’s bundle , and (1 l) is the exponent on his own bundle, the opposite of where these terms appear in Ayanda’s utility function. The totally self-regarding person, Biko in this case, places no weight on the bundle of the other person; his degree of altruism, l = 0. So self-regarding B’s utility function is: uB (xA , yA , xB , yB ) = = ⇣ (1 xAa yA ⌘ ⇣ ⌘ a) 0 (1 a ) 1 xBa yB (1 a ) xBa yB (4.18) which is just his previous utility function before we introduced l . The rearrangement of the equation in the second line is true because any term raised to a zero exponent (as in Biko’s utility function) has a value of 1. Checkpoint 4.8: Spite and love . a. What would it mean in the utility function 4.16 if we had l < 0? Can you give an example of someone acting as if they had preferences like this? b. Can you imagine a person having a value of l greater than one half, what would this mean? Can you think of situations in which people have acted on preferences of this type? An altruistic indifference map To draw her indifference map, we will give Ayanda some particular value of l . Figure 4.14 shows an Edgeworth box representing a-not-perfectly-altruistic Ayanda with l = 0.4. Ayanda’s indifference curves look like the contours on a topographic map of a mountain. We described the constrained optimization process in Chapter 3 as a kind of hill climbing. where both elements in the bundle were a "good" and over the entire map, the mountain rose to higher levels if you moved in the "north east" direction, that is more of both goods. In those figures you never saw the top of the mountain, because there was not any top. There was no such thing as "too much" of either good. But Ayanda’s indifference map has a definite peak at the allocation indicated by point v. The reason is that from her other-regarding perspective she can have "too much" of a good when that means that Biko (who she cares about) too little. This is why Ayanda’s indifference curves oval shaped, just like the description of a mountain and its peak on a contour map. 213 MICROECONOMICS 9 B's coffee (kilograms), xB 7 6 5 4 3 2 Biko 1 0 k A's highest utility v uA 4 uA3 uA2 uA1 j z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 9 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 8 B's coffee (kilograms), xB 7 6 5 4 3 2 0 k A's highest utility uB1 Biko 1 uB2 uB3 v uA 4 uB4 uA3 uA2 uA1 i Pareto−efficient Curve j z uB5 0 Ayanda 1 (a) Altruistic Ayanda’s indifference curves 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 B's data (gigabytes), yB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 8 B's data (gigabytes), yB A's data (gigabytes), yA 10 - DRAFT A's data (gigabytes), yA 214 10 (b) Altruistic Ayanda and self-regarding Biko Notice that when she has little of either good (close to her origin in the lower left of the box) her indifference curves look as you have seen before. In this situation both coffee and gigabytes are "goods" so more of each is better, and the indifference curves slope downward, as you would expect. Moving up or to the right brings you to a higher indifference curve. In this part of the figure "more is up." But beyond a certain point "more" for Ayanda is no longer "up". If she has most of both goods then getting even more is not something she values, so moving up and to the right leads her to lower not higher indifference curves. To understand the upward-sloping parts of Ayanda’s indifference curves, remember that if one of the axes represents a good and the other a bad, then the indifference curve slopes upwards, as in the case of study time (a bad) and expected grades (a good). In the upper right of the box for example near point k where she has most of both goods and Biko has little of either the indifference curves slope downward because for Ayanda having more of either good (and Biko having less) reduces her utility: both her coffee and her gigabytes are "bads" not goods. In panel b of Figure ?? we add Biko’s conventional (self-regarding) indifference curve, so we now know how both of them evaluate every feasible allocation given by the dimensions of the box. To do this we use Biko’s selfregarding utility function with the value he places on Ayanda’s utility being zero that is l = 0 because he is entirely self-regarding (that is, zero altruism). The Pareto-efficient curve is, as before, made of points of tangency between Ayanda’s and Biko’s indifference curves. But now we exclude tangencies at allocations for which Ayanda places a negative value on having more of one or Figure 4.14: Allocation and distribution with one altruistic person and one self-regarding person. In panel a the green oval shaped curves labeled uA are the indifference curves based on Ayanda’s utility function. In both panels, points z and i are the same allocations here as in Figure 4.6. Notice that in panel a because Ayanda values what Biko gets she regards the j as equivalent to the endowment k, despite the fact that she receives less of both goods at j than she does at k. For the same reason, Ayanda’s utility reaches a maximum at the allocation v indicated in the figure. The Pareto-efficient curve now does not include k, because Biko is so deprived of both goods at that point that Ayanda prefers v to k. P R O P E RT Y, P OW E R , 8 7 6 5 4 3 2 Biko 1 0 uB1 uB2 uB3 A v uB4 uA3 uA4 uA2 uA1 i vB Pareto−efficient curve z 0 Ayanda 1 2 3 4 5 6 7 A's coffee (kilograms), xA 8 9 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 9 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 B's coffee (kilograms), xB 8 7 6 5 4 3 2 1 Biko 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 uB1 uB2 vA uB3 uA5 Pareto−efficient curve uB4 uA4 i uA3 uB5 vB uA2 uA1 0 Ayanda 1 2 (a) More altruism (l = 0.4) both of the goods, above and to the right of her "utility peak" at v. As a result the Pareto-efficient curve in Figure 4.14 looks different from the one in Figure 4.6 as it does not extend upwards and to the right beyond Ayanda’s maximum v. Ayanda does not want more of either good than she gets at her maximum v, while Biko prefers j to any allocation in which she gets less of either or both of the goods. Checkpoint 4.9: Altruistic comparisons Consider Figure 4.14 a. Where is Biko’s utility peak in the figure (analogous to Ayanda’s allocation at point v? b. where would point v be if l = 12 (or as close to l = 12 as possible? c. What happens if Ayanda is self-regarding and Biko is an altruist? How would the Edgeworth box change? Efficiency and fairness among altruists With these analytical tools we can now look at the decision problem faced by the friends Ayanda and Biko both with other-regarding social preferences. Figure 4.15 hows for the same Edgeworth box, the indifference maps of the two. Unlike the case of one altruistic actor, now both participants have preferred allocations in the interior of the Edgeworth box. They both would like to avoid "too much of a good thing." Both of them dislike extreme allocations giving most to one or the other. This would not be the case were they evaluating bundles in isolation, that is if the other person did not exist. The reason why they place a negative 3 4 5 6 7 A's coffee (kilograms), xA 8 215 z 9 B's data (gigabytes), yB 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 B's coffee (kilograms), xB A's data (gigabytes), yA 9 B's data (gigabytes), yB A's data (gigabytes), yA 10 & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 10 (b) Less altruism (l = 0.2) Figure 4.15: Altruistic indifference maps. The two panels depict two different levels of altruism: high (l = .4) in panel a and low (l = .1) in panel b. The allocations indicated by the points vA and vB are respectively A’s and B’s preferred allocation. The Pareto-efficient curve is composed of all allocations at which both own coffee and own data are "goods" rather than "bads" to both A and B, and where their marginal rates of substitution are equal, that is, their indifference curves are tangent. 216 MICROECONOMICS - DRAFT value on getting more when they already have a lot is not due to diminishing marginal utility, it is because getting more for yourself means getting less for the other. Each of their preferred allocations are shown in the figures by the allocations, vA for Ayanda and vB for Biko. Around each person’s preferred allocation, their iso-social welfare curves move outwards and downwards in all directions, corresponding to lower and lower levels of utility. As you can see from panel a of Figure 4.15 the Pareo efficiency curve is a line between their two preferred "utility peaks" vA and vB . By comparing panels a and b depicting greater and lesser degrees of altruism, you can see that the more altruistic they are, the shorter the Pareto-efficient curve is, because greater altruism eliminates more of the extremely unequal allocations. There is still a conflict of interest, however. At Ayanda’s preferred allocation Biko has a level of utility less that than the utility he enjoys at this own preferred allocation. The same is true of Ayanda: she does much better at her preferred allocation than at Biko’s Along the Pareto efficient curve movements in one direction or the other necessarily involve one gaining and the other losing. As always the Pareto efficient curve is a conflict region even among altruists. The fact that the ’utility peaks’ are closer together in panel a illustrating a greater degree of altruism means that the conflict of interest between them is lesser the more altruistic they are. How might they resolve their remaining conflicts of interest? Here, to make a decision, they need to go beyond their own utilities (even taking account of their altruistic nature) to bring in some additional way of making a judgement. They might adopt: • a social norm that they both share, for example if one of the two found the coffee and the data they could go by "finders keepers"; in this case whichever of them who found the goods could make the decision, presumably implementing his or her preferred allocation. • a procedural rule of justice, for example flipping a coin to see whose preferred allocation vA or vB would be implemented; or • a substantive rule of justice, for implementing the allocation recommended by the Impartial Spectator. Point i in the figures is a reference point showing the allocation that the Impartial Spectator (who weights Ayanda’s and Biko’s utilities equally) would implement. This is the same allocation that they would have both preferred had they been perfect altruists. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 217 Checkpoint 4.10: Altruism and Rents Why does altruism reduce the conflict over which allocation to implement? 4.13 The rules of the game and the problem of limited information Case Constraints implied by the rules of the game Objectives of the actor(s) Characteristics of the resulting allocation Impartial spectator The available goods (dimensions of the Edgeworth box) Social welfare equally weighting the utility of each Pareto efficient and fair (by the standard of the social welfare function). Symmetrical bargaining with no first-mover advantage Endowment allocation (private property). Each player’s participation constraint (PC) at each stage of the bargaining Utility of the two traders Pareto-efficient if no impediments to bargaining, otherwise possible Pareto improvements over the endowment allocation Take-it-or-leave-it power Endowment allocation (private property) Second mover’s PC Utility of the first and second movers Pareto efficient, first mover’s rent is all of the gains from exchange Price-setting Power Endowment allocation (private property) Second mover’s incentive compatibility constraint (ICC) Utility of the first and second movers Pareto inefficient; first mover gets most of the gains from exchange but 2nd mover gets some Legislation The available goods (money and time) Whatever the legislators were seeking to (possibly the social welfare optimum) Pareto inefficient, could be improved upon by private bargaining Bargaining away from legislated hours and wages The new participation constraints given the fallback position implemented by the legislation Utilities of the two players Pareto-efficient if no impediments to bargaining, otherwise possible Pareto improvements over the new fallback Altruism The available goods (dimensions of the Edgeworth box) Utilities of both (taking account of how much they value the other’s bundle); fairness Pareto efficient and (if they can agree on a fairness principle) fair. We have examined several institutional approaches to resolving the conflict between Ayanda and Biko over allocations of available goods. They all illustrate the dilemma posed in social interactions between: • The goal of reaching an allocation that is Pareto-superior to the endowment and possibly even Pareto-efficient. • The goal of resolving the conflict over the distribution of the resulting eco- Table 4.1: The rules of the game: Objectives, constraints and the characteristics of the resulting allocations. 218 MICROECONOMICS - DRAFT nomic rents in a way that is fair. Table 4.1 summarizes some of the key aspects of the cases we have discussed. Which of the scenarios in the table are relevant in any particular case depends on the rules of the game for the society of which the players are a part. How well the rules work depends in important part on whether the actors have the information that we have attributed to them. • The Impartial Spectator needs to know a lot about Biko and Ayanda to implement his preferred outcome, in particular their preferences. • The symmetrical traders need little information other than their own preferences; they simply continue accepting exchanges as long as the post exchange bundle is preferable to the pre-exchange bundle. • The person with TIOLI power needs to know the second-mover’s participation constraint (a single indifference curve), which is less information than the Impartial Spectator requires. • The person with price-setting power needs to to estimate the secondmover’s best-response incentive compatibility constraint, which requires more information than the participation constraint, but less information than the Impartial Spectator. If the legislator (who imposed the hours and wages law) was intending to implement an efficient and fair outcome such as the one recommended by the Impartial Spectator, then he (or they) would have to know as much as was required of the Impartial Spectator, namely the entire preference maps of the two. • The two altruists need to know both their own and the others preferences (without knowing what the other cares about it is impossible to care about the other). This is as demanding as the information required of the Impartial Spectator. A basic fact of economic life is that information is scarce and local. For example in their altruistic friends scenario Ayanda and Biko probably know a lot about each others preferences, but this is unlikely when Ayanda is the employer and Biko her prospective worker. This will have important ramifications in the chapters to come especially when we study the labor market, the credit market, and other exchanges where limited information makes it impossible to implement Pareto eficient allocations. 4.14 Conclusion From the silent trade that Ibn Battuta and Herodotus described centuries ago to eBay, Amazon and Alibaba today, people have exchanged goods to their mutual advantage and engaged in conflicts over who would get the lion’s share of the the gains from exchange. The four scenarios we have H I S TO RY For Friedrich Hayek, an important 20th century economist and philosopher, the fact that information is scarce and local was the basis of his criticism of centrally planned economies – such as the Soviet Union at the time – and his advocacy of private property and markets. See his thoughts on this in the headquote for Part IV of this book and in Chapter 14. P R O P E RT Y, P OW E R , & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S introduced have made it clear that the outcomes of these exchanges and conflicts depend on the institutions under which they take place, and the preferences of the people involved. We will see in later chapters that it is quite common that one of players in an economic interaction has price-setting power or its equivalents: the power to set wages, interest rates, and other terms of an exchange. We have also seen some of the scenarios require that an actor knows a lot about the other person which is not very realistic even in the two person case we have used as a simplification of societal interactions. The fact that information is both scarce and local will play an important role in our analysis of capitalism as an economic system in later chapters. The abstract scenarios we have introduced here do not capture the often stepby-step dynamics by which people move from their endowments to eventually reach an allocation through a series mutually beneficial trades. We will see that exactly how an economy moves from an out-of-equilibrium endowment to an equilibrium final allocation makes a difference for the fairness of the outcomes, but to economists it remains a vexing and far from settled problem. Making connections Constrained optimization in strategic interactions: The constrained optimization techniques developed in Chapter 3) are used to better understand strategic interactions introduced in Chapters 2 and 1. Optimization rules. In addition to the mrs = mrt rule which we developed in Chapter 3 for individual optimization we also have the mrsA = mrsB rule defining a Pareto efficient outcome, both of which are used in strategic interactions. Mutual gains from trade: If the endowment allocation (status quo) is not Pareto-efficient, then mutual gains are possible by implementing some different allocation of the goods which people may be able to agree to voluntarily. Rents and conflicts: These improvements over the fallback option accruing to the players are rents, made possible by the gains from exchange. Institutions (rules of the game) and bargaining power: The distribution of these rents in the Nash equilibrium allocation depend on the players preferences and the initial endowment as well as on the property rights in force, other institutions and the forms of bargaining power that each participant can exercise. Pareto efficiency, institutions: If players have sufficient information some insti- 219 220 - DRAFT MICROECONOMICS tutions will result in Pareto-efficient outcomes. Examples are the allocation implemented by the imaginary Impartial Spectator, and the situation in which one person has take-it-or-leave-it power. Price-setting power by one person, however, results in a Pareto-inefficient outcome even with unlimited information. Self-regarding and social preferences: Among the set of Pareto-efficient allocations there will generally be conflict of interest among the participants. But, the extent of these conflicts may be reduced by social preferences such as altruism. Important Ideas utility function marginal rate of substitution Cobb-Douglas Utility Edgeworth box pareto-criterion pareto-improving lens pareto-efficiency pareto-efficient curve utility possibilities frontier endowment post-exchange allocation impartial Spectator social welfare function mrsA = mrsB rule iso-welfare curve altruism private property first-mover advantage take-it-or-leave-it power price-setting power participation constraint incentive compatibility constraint price-offer curve institutions gains from trade economic rent Mathematical Notation Notation Definition a u() x̄, ȳ p W l h() a exponent of good x in the Cobb-Douglas utility function utility function total amounts of x and y available for trade price of coffee in terms of data Social Welfare function extent of altruism non-linear term of quasi-linear utility function parameter in the linear term of quasi-linear utility function Note on super- and subscripts: A, B and i: people; z: endowment point; ti : outcome with a take-it-or-leave-it offer. Discussion Questions See supplementary materials. P R O P E RT Y, P OW E R , Problems See supplementary materials. Works cited & E X C H A N G E : M U T UA L G A I N S & C O N F L I C T S 221 5 Coordination Failures & Institutional Responses DOING ECONOMICS This chapter will enable you to do the following: Right now, my only incentive is to go out and kill as many fish as I can...any fish I leave is just going to be picked by the next guy. 2 John Sorlien. Rhode Island (USA) lobsterman Don’t get him wrong: John Sorlien, the lobsterman, is not the kind of selfinterested and amoral Homo economicus you might find in an economics textbook. He is actually an environmentalist of sorts, and as President of the Rhode Island Lobstermen’s Association he was up against a serious problem of incentives, not a shortcoming of human nature. When he started lobstering at the age of 22, he set his traps right outside the harbor at Point Judith, within a few miles of the beach, and made a good living. But the inshore fisheries have long since been depleted, and now his traps lie 70 miles offshore. He and his fellow lobstermen are struggling to make ends meet. Across the world in Port Lincoln on Australia’s south coast, Daryl Spencer, who dropped out of school when he was 15 and eventually drifted into lobstering, has done much better. During the 1960’s the Australian government assigned licenses – one per trap – to lobstermen working at the time, and from that time on, any newcomer seeking to make a living trapping lobsters off of Port Lincoln had to purchase licenses. • Understand how the external effects of our actions on others that are not taken into account when people make choices lead to coordination failures. • Represent social interactions with graphical and algebraic indifference curves, feasible sets and best responses functions. • Use best response functions to see how the fairness and Pareto efficiency of the resulting allocations will depend on the rules of the game. • Understand how improvements in property rights, government policies such as taxes or direct regulation, the exercise of power by private individuals and social preferences can all result in Nash equilibria that are Pareto-superior to the Nash equilibrium that would result in their absence. • See that the Pareto-improvement made possible in each of these cases occurs because (in very different ways) they induce actors to internalize the external effects that their actions have on others. Spencer purchased his start up licenses for a modest sum and by 2000 his licenses were worth more than a million U.S. dollars (in 2000 prices); considerably more valuable than his boat. More than giving Spencer a valuable asset, the policy has limited the Australian lobstermen’ s work: Spencer has 60 traps, the maximum allowed; in the Atlantic off of Point Judith John Sorlien pulls 800 traps and makes a lot less money. Regulating the amount of lobsters trapped is a coordination problem. Point Judith and Port Lincoln represent extremes along a continuum of failure and success; with the lobstermen of Port Lincoln reaping the mutual gains made Figure 5.1: Sounding the alarm on climate change, a coordination problem. Greta Thunberg, then 16 years old, speaking at the United Nations in 2019 about what is probably the most serious coordination problem that humanity has ever faced. She said: “We are in the beginning of a mass extinction, and all you can talk about is money and fairy tales of eternal economic growth. How dare you!”1 224 MICROECONOMICS - DRAFT possible by a joint decision to limit the number of traps. One may wonder why the Point Judith fishermen do not simply emulate the Australians. This is especially surprising since one of Sorlien’s friends and a fellow Point Judith lobsterman visited Port Lincoln, returning with tales of millionaire fishermen living in mansions. But getting the rules right is a lot more difficult than the Port Lincoln story may suggest, and good rules often do not travel well. One of the common obstacles to successful coordination is that the rules that address the coordination problem also implement a division of the gains to cooperation. In Port Lincoln, those who were awarded the licences benefited; others did not. Had the young Daryl Spencer not agreed one day to help out a lobsterman friend and then decided to become a lobsterman himself, someone else would be a millionaire, and Spencer might still be painting houses and complaining about the high price of lobsters. Even if policies to address coordination failures could result in benefits for everyone affected, how a group coordinates, and what policies they coordinate on will affect how these benefits will be distributed. And this makes it difficult to agree on a policy. An example is the Ultimatum Game experiment, in which conflicts over the size of the Proposers and Respondent’s "slice of the pie" sometimes result in neither getting any piece of the pie at all. Conflicts over the distribution of the gains to cooperation have sunk many otherwise viable agreements to limit the depletion of fishing stocks. To give an example, a confederation of tribes of North-west U.S. Native American salmon R E M I N D E R A coordination problem is a situation in which people could all be better off (or at least some be better of and none be worse off) if they jointly decide how to act – that is, if they coordinate their actions – than if they act independently. fishermen seeking to limit their catch decided to allocate shares of a given maximum catch to each tribe. In the course of months of debate and bargaining the following principles of division were proposed, with each proposal more or less transparently benefiting one or another tribe or class of people:3 • One tribe one share. • Shares allocated in proportion to a tribe’s number of members. • Shares to each tribe based on the tribe’s expenditure on lobbying (seeking to influence) the U.S. Federal Government to adopt policies more favorable to the tribes, and finally. • Shares to each tribe in proportion to the relative quantities of fish taken at the time of the initial treaty with the U.S. Government. Neither unrestricted competition nor marketable permits to catch specified amounts (similar to Australia’s lobstering licenses) was proposed. The variety of proposals and their different effects on the distribution of income among the tribes suggest how challenging it may be to agree on a rule for sharing the gains to cooperation. P ERMIT A permit allows a firm or person to engage in an activity: it gives them permission. A permit gives the holder a property right to a certain amount of a good or output. For example, a fishing permit would allow a certain number of fish to be caught or a carbon emission permit would allow a certain amount of carbon dioxide to be emitted during production. When permits are transferable, firms and people can buy and sell permits at a price. C O O R D I N AT I O N F A I L U R E S Depleting a fishing stock is little different in the structure of its incentives and its consequences from many other social interactions. In Chapter 9, for example, using exactly the model we develop here of the coordination problem that fishers face "harvesting fish," we will study how firms compete on markets "harvesting customers" by attempting to charge lower prices than their competitors. What do firms competing on markets have in common with fishing people depleting the basis of their livelihood? The common idea is over-harvesting – whether it is fish or customers – that could be prevented if the firms or fishermen coordinated their actions rather than acting singly. Just as the Port Lincoln lobstermen discovered that they could benefit by making a common decision to limit the number of traps they set, so too will firms discover that they could make higher profits if they were able to agree on a price at which to sell, rather than competing. In Chapter 9 we will return to the fact that coordination among the firms to set a common price – which is illegal in many countries – raises profits but harms buyers and as a result increases inequality. The fact that coordination problems take so many familiar forms explains both the continuing interest in Hardin’s “Tragedy of the Commons” introduced in Chapter 1 as well as the impressive amount of human ingenuity that has been invested in finding ways to avoid or mitigate the costly consequences of uncoordinated individual optimization in these situations. In this chapter we develop tools to understand the nature of coordination problems like the Tragedy of the Commons. We use these tools to analyse some of the policies (changes in the rules of the game) that improve the Nash equilibrium outcome when external effects are present. 5.1 Common property resources, public goods, and club goods Coordination problems are common because when we interact with others we affect their well being – positively or negatively – and these external effects are not taken into account when we decide on a course of action. The nature of these external effects differs depending on the type of interaction in question. In the case of over fishing or ‘over-harvesting’ consumers, when one person fishes more, or a firm cuts prices, the external effects – on the catch of the other fishermen or the profits of other firms – are negative. A taxonomy of goods To better understand the kinds of coordination problem that we face and how we might design effective remedies, we classify goods according to their the & INSTITUTIONAL RESPONSES 225 226 MICROECONOMICS - DRAFT kinds of external effects associated with them and the reason why these are a problem. To do this we ask two questions, introducing two new terms: • Is the good rival or non-rival? • Is the good excludable or non-excludable? When a good is rival, the benefits of its use are limited: more people using the good reduces the benefit available to others. Your phone is a rival good (our using it precludes others using it at the same time) while information typically is non-rival (the fact that I know what time it is and share this information with you does not preclude your benefiting from the same information). The distinction between rival and non rival goods can be dramatized by considering how different the reaction would be if you met someone in the street and politely asked: • "Excuse me could you give me the time of day?" or • "Excuse me, could you give me your phone? When a good is excludable a potential user may be denied access to the good (or excluded from its usage) at low cost. Your home is an excludable good. The music from an outdoor concert in a park is not excludable. We make use of these distinctions to provide the taxonomy shown in Table 5.1. The four categories shown there are "pure cases" introduced to clarify distinctions. In reality many goods or resources have some aspects of a public good (they may be a little bit rival and a little bit excludable). The same is true of the other three categories. Non-excludability and external effects But if we just think about the pure cases for now, we have the following: Common property resources are rival and non-excludable, like in the Fishermen’s Dilemma in Chapter 1. As was the case for the lobstermen above, the more one fished, the less others caught; but in the absence of a permit system like they adopted in Australia, no fishermen could be stopped from fishing, so the common property or pool (the lake or the ocean) was nonexcludable. Examples of common property resources and their associated coordination problems include congestion in transportation and communications networks, overuse of open access forests, fisheries, water resources. Status is another common property resource, not everyone can be high status (there is a limited amount to go around) so it is rival. But nobody can be excluded from acquiring status symbols and engaging in other social climbing activities. C O O R D I N AT I O N F A I L U R E S Rival Non-rival & INSTITUTIONAL RESPONSES Excludable Non-excludable Private good Common Property (Pool) Resource (clothing, food) (fishing stocks, potential buyers), Club Good Public Good (streaming music, online movies) (global climate, rules of calculus) A public good is both non-rival and non-excludable. A private good is neither: it is both rival and excludable. A slice of pizza is a private good: it is rival because if you eat it nobody else can enjoy it. It is excludable because the pizza seller can exclude you from eating it if you do not pay for it. By contrast, weather forecasts (on your phone, website, or the radio) are a public good. As more people use the weather forecasts the benefits that those already using the forecasts receive do not decrease, the benefits of the weather forecasts are non-rival. No person can be excluded from access to the information about the weather, therefore the benefits are non-excludable. When a person contributes to a public good – for example by producing some new information of value to everyone – she is contributing benefits to others, so she confers external benefits on others. The problem here is that the person does not benefit from the positive external effects that her actions convey on others. So unless the actor values the well being of others as much as hers own (very unlikely) the public good will be under-provided. Common pool resources, as the lobsterman John Sorlien explained, will be over-exploited. In contrast with public goods and common property resources, there are "club goods." Club goods are non-rival, but people can be excluded from their consumption. Common examples include collecting a toll on a little used highway, charging admission to an uncrowded museum, or making people pay for streaming video and music. Intellectual property rights such as patents and copyrights are club goods: allow people to be excluded from the use of information, which in the absence of the intellectual property rights would be a public good. This makes it clear that how some good or resource is classified in our two-by-two taxonomy depends not only the nature of the good itself, but also on the rules of the game that determine whether it is excludable or not. In this chapter we illustrate how coordination failures occur and how policies might address them with the example of common property resource problems (or common pool resource problems). The "common property" or "common pool" is the stock of fish available for catching or the pool of customers who might purchase the goods sold by the firms. Because common property resources are non-excludable and rival, people who use them impose external 227 Table 5.1: Public, private, common property and club goods. In parentheses are examples of the kinds of goods. 228 MICROECONOMICS - DRAFT Figure 5.2: he Grand Banks (North Atlantic) fisheries: cod landings in tons (1851-2014). In the 1960s new fishing technologies allowed a dramatic increase in cod fish caught ("landings") far outpacing the capacity of the fish to reproduce. This led to a partial collapse of the fishery in the 1970s and a total collapse in 1992 when the Canadian government banned fishing entirely. The fishing stocks to sustainable levels of the past by the 2030s. Source: Ecosystems and Human Well-being: Synthesis. A Report of the Millennium Ecosystem Assessment (2005)4 Landings (tons of cod) 600000 400000 200000 0 1850 1870 1890 1910 1930 1950 1970 1990 2010 Year costs on each other. The "problem" is that self-regarding people will overexploit the resource because they will not place any value on the negative external effects of their actions on others. Just such a pattern of exploitation is shown in Figure 5.2, which displays the catches of cod fish in the North Atlantic fisheries. Checkpoint 5.1: A taxonomy of types of goods Look again at Table 5.1 think of at least two further examples for each of the four categories of goods. 5.2 A common property resources problem: Preferences Let’s consider a specific example of a common property resource problem: the over-exploitation of an environmental resource. It could be the oceans, or forests, or a livable planet, but we’ll stick to the problem of over-harvesting fish. Our questions look at the ways that the rules of the game and the preferences of the actors determine what we should expect to happen in these situations. Preferences over fishing time and fish consumed We turn now to the the problem confronted by two fishermen, called Abdul (A) and Bridget (B). We model just two fishermen as a way of representing C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 229 Figure 5.3: Abdul and Bridget trying to catch the same fish. The lake is a common pool resource, so the benefits are rival and each imposes a negative external effect on the other. how a large number of them might interact. They fish in the same lake, using their labor and their nets. To start, we assume they consume the fish they catch (what we call their "catch") and do not engage in any kind of exchange. We will begin by assuming that they do not make any agreements about how to pursue their economic activities. (Recall that this means that they are engaged in a non-cooperative game.) Each derives well-being from eating fish and experiences a loss of well-being (disutility) with additional fishing time. We represent their preferences when they are engaged in some amount of fishing with the following quasi-linear utility functions: Fisherman’s utility Abdul’s utility Bridget’s utility = Fish consumption 1 A 2 uA (hA , yA ) = yA (h ) 2 1 B 2 uB (hB , yB ) = yB (h ) 2 Disutility of fishing (5.1) (5.2) The utility function given by Equation 5.1 tells us three things about Abdul’s preferences: • Consumption (yA ) measured in pounds of fish is a “good;" Abdul derives utility from obtaining more consumption (consuming more fish) which is why yA has a positive sign. • Time spent fishing (hA ) measured in hours is a “bad": the second term has a negative sign. • Utility (uA ) is increased by one unit if he is able to consume one more pound of fish, so the units in which we can measure utility are pounds of fish. • Marginal utility is not diminishing but instead is a constant (equal to 1, D ISUTILITY OF WORK Working doesn’t only take up time, it is also costly to people because of the effort that they need to exert. Manual labor is physically tiring and often, with activities like construction and mining, can be dangerous as well as complex and challenging mentally. Working as a waitress burns as many calories in an hour as doing construction work. Office work, too, requires effort, requiring concentration and attention. Exerting this effort often isn’t pleasant and therefore results in disutility or a cost of utility to exert. 230 MICROECONOMICS - DRAFT uA3 uA2 Consumption, yA slope = hA = 15 Figure 5.4: Abdul’s indifference curves over output (yA ) and fishing time measured in hours (hA ). Output (fish) (yA ) is a “good" and provides Abdul with positive utility, whereas fishing time (hA ) is a “bad". Notice that Abdul’s indifference curves in fishing hours and output are upward-sloping, similar to the indifference curves over money (income, a good) and working time (a bad) in Chapter 4. uA1 slope = hA = 10 uAz g yg = 247.5 f yf = 190 z = 112 0 5 hAf = 10 hAg = 15 20 24 A's hours, hA because an additional pound of fish provides him with a one unit increase in utility). Bridget’s utility function Equation 5.2 is interpreted in the same way as Abdul’s. Both of them refer to some given time period, such as a week. So output and consumption are pounds of fish caught and eaten in a week, while time spent fishing is hours fished over the course of a week. If for some reason they do not fish at all, they receive a small amount of fish yz from neighbors or the government, labeled with the subscript z because this is their fallback position (as the endowment allocation was in Chapter 10) To decide how much time to fish, people like Abdul have to balance their disutility of hours of work with the utility of consumption that they get from consuming the fruits – or the fish – of their work time. To understand this process, we look at Abdul’s indifference curves. Four indifference curves derived from Abdul’s utility function, equation 5.1, are presented in Figure 5.4. Notice that: • the higher numbered (meaning more preferred) indifference curves are above (more fish) and to the left (less work); • the curves slope upwards because fish is a good and fishing time is a bad, so comparing points f and g he is indifferent between fishing less and consuming less (point f) and fishing more and consuming more (point g) R E M I N D E R The indifference map provides information on how he evaluates all of the imaginable combinations of fishing time and fish caught. It says nothing about the actions and outcomes that are feasible for him. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 231 • the lowest indifference curve is labeled uA z and its vertical axis intercept is point z or the level of utility measured in fish per week, yA z that he will receive if he does not fish at all; and finally • for any given level of yA the indifference curve is steeper the more hours Abdul works: the more he works the greater is his dislike of working more compared to how much he likes eating more fish. The negative of the slope of his indifference curve is Abdul’s marginal rate of substitution between fish (yA ) and fishing time (hA ), the ratio of his marginal utility of fishing time to his marginal utility of fish. This quantity takes a particularly simple form in this case because (as is shown in M-Note 5.1). Abdul’s marginal utility of fish is 1 and his marginal utility of fishing time is hA . So, the marginal rate of substitution of fish consumption for fishing time is: mrsA (hA , yA ) hA = (5.3) Abdul’s marginal rate of substitution of fish consumption for fishing time is hA , and this is also his marginal disutility of fishing time, which is negative, because he regards fishing time as a “bad.” If he were already working 12 hours, then his disutility of hours of fishing (which is hA itself) is the greatest amount of fish he would be willing to give up in order to be able to work an hour less. This is his willingness to pay (in fish) to have more free time. Or if Abdul were employed where he is already working hA and paid a wage, then the quantity hA is the lowest wage (paid in fish) in return for agreeing to work an extra hour, that he would accept. M-Note 5.1: The mrs(h, y) with quasi-linear preferences When Abdul’s utility is given by Equation 5.1, we have. Marginal utility of consuming fish = ∂ uA (hA , yA ) =1 ∂ yA (5.4) Marginal disutility of fishing time = ∂ uA (hA , yA ) = hA ∂ hA (5.5) The marginal utility of fishing time is negative (it reduces Abdul’s utility and is equal to hA ) and we use the term marginal disutility of fishing time for the same quantity but with a positive sign (it increases Abdul’s disutility). The marginal rate of substitution of output for hours of work (mrsA (hA , yA )) is the negative of the slope of the indifference curve, which is ratio of the marginal utilities: mrsA (hA , yA ) = hA = 1 hA (5.6) So the slope of the indifference curve is the marginal disutility of working time, or just hA the amount of working time itself. A A Furthermore, along any of the indifference curves, uA 1 , u2 and u3 , the vertical intercept is the amount of utility in fish if they were not working at all, that would be the same as the utility at every point on that indifference curve. The marginal rate of substitution of Abudl’s fish consumption for Abdul fishing time M - C H E C K Abdul’s utility function in fish and fishing time is quasi-linear : since it is linear in fish – he derives a positive and constant marginal utility from consuming fish – but is negative (it is a disutility) and non-linear in fishing time. His marginal disutility of fishing time it is not constant, it is greater the more time he spends fishing. 232 MICROECONOMICS - DRAFT mrs(hA , yA ) is an entirely different quantity than the marginal rate of substitution given by indifference curves for the fishing times of the fishermen mrs(yA , yB ) that we introduce later. Checkpoint 5.2: The lake as a common property resource a. Explain why the lake that Abdul and Bridget are fishing is a common property resource. What are its characteristics? Explain. b. Return to Chapter 1 and the choice of strategies that the fishermen had in the Fishermen’s Dilemma to Fish 10 hours or Fish 12 hours. Substitute these values into the utility functions to see what the payoffs in the corresponding game table would be if the fishermen could only choose these two strategies. Find the Nash equilibrium of the game. 5.3 Technology and environmental limits: The source of a coordination failure A coordination problem arises because Abdul or Bridget fishing more reduces the amount of fish the other catches in an hour of fishing. This negative T ECHNOLOGY A technology is a description of the relationship between inputs –including work, machinery, and raw materials – and outputs. external effect that each has on the catch of the other is the source of the coordination problem. These external effects are part of the technology of fishing. A technology is a description of the relationship between inputs – such as fishing time, equipment, and fish in the wild – and outputs – in this case caught fish. A technology is often described mathematically in a production function. You already used a production function in Figure 3.9 where the input was time spent studying and the output as learning. (We postpone a detailed discussion of production functions until the next chapter). ?? Here are the production functions for Abdul and Bridget, where xA for Abdul and xB for Bridget represents the number of fish caught by each of them in a week and hA and hB are the hours of fishing time they work during the week. The production functions translate the actions taken by the two – their fishing hours (hA and hB ) – into the amount that each catches (xA and xB ) and consumes (yA and yB ). Abdul’s catch: xA (hA , hB ) = hA (a b (hA + hB )) (5.7) Bridget’s catch: xB (hA , hB ) = hB (a b (hA + hB )) (5.8) The two parameters of the production function are: • a (Greek alpha) is the the fisherman’s maximum average productivity, that is, total catch divided by time spent fishing which would occur if one of P RODUCTION F UNCTION A technology is a way of transforming inputs into outputs, described mathematically as a production function. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 233 them fished some small amount of time and the other did not fish at all. We assume that a > 0 otherwise they could not ever catch any fish. • b (Greek beta) measures the decrease in average productivity for each hour fished in total by the two. We assume that b > 0 to reflect the fact their interdependence and the negative external effect that each fishing has on the other’s catch. Were we to be applying this model to real fishermen, we would find that b and a differ. For example, one may have a larger boat and for this reason may catch more fish in an hour and also have a larger effect on the fishing productivity of the other. However, because our interest here is not in the effects of differing sizes of their boats, but instead on differing amount of time fished, we assume that b and a are the same for the two fishermen. From Equation 5.7 you can see that Abdul’s total catch is his hours of fishing multiplied by his total catch per hour of fishing, termed the average productivity of his fishing time. Dividing both sides of Equation 5.7 by hA we have his average productivity of fishing time: xA hA = a b (hA + hB ) (5.9) M - C H E C K We adopt parameters for the production functions so that Bridget and Abdul cannot work so many hours that their average productivity becomes negative, so that fishing more would reduce their total catch. This is why we do not extend the lower of the two production function curves in Figure 5.5 beyond 24 hours, the point after which the function turns downwards. What this means is that: Average productivity = Maximum Reductions due to own and other’s fishing time We are also interested in what is termed the marginal productivity of Abdul’s fishing time. This is the effect of fishing a little more on the size of his total catch. In the M-check we show that Abdul’s marginal productivity of fish time is: mphA = a b (hA + hB ) b hA (5.10) This equation can be read as: Marginal productivity = Average productivity Reduction due to own fishing time We call a the maximum productivity of fishing because it is the amount that would be produced per unit of fishing time when there is no fishing being done. The parameter b expresses three important aspects of the technology: • Decreasing average productivity: If Abdul spends more time fishing, his catch will be larger, but his average productivity – the size of the catch per hour fished – decreases. • Decreasing marginal product of work time : If Abdul already fishes a lot, then the additional amount of fish that he catches were he to fish a little more will be less than if he were initially fishing a lesser amount. M - C H E C K The marginal product of Abdul’s fishing time is found by partial differentiating his total catch (xA ) given by the production function (Equation 5.7) with respect to his working time (hA ), which gives us Equation 5.10 We study the mathematical and conceptual properties of production functions and marginal products in the next chapter. MICROECONOMICS Consumption, yA (pounds, lb) 234 - DRAFT Figure 5.5: Abdul’s production of fish with hours of fishing and marginal benefit of hours spent fishing. In the top panel, Abdul’s light green total product line corresponds to when Bridget does not fish (hB = 0) and Abdul’s light green total product line corresponds to when Bridget fishes 12 hours (hB = 12). Similarly, in the lower panel, Abdul’s light green marginal benefit line corresponds to when Bridget does not fish (hB = 0) and Abdul’s light green marginal benefit line corresponds to when Bridget fishes 12 hours (hB = 12). y(hA, hB = 0) k 288 y(hA, hB = 12) j 216 0 3 6 9 12 15 18 21 24 27 30 27 30 A's marginal benefit (pounds, lb) 30 24 k 18 Marginal benefit when, hB = 12 0 3 6 Marginal benefit when, hB = 0 j 12 9 12 15 18 21 24 A's hours, hA • Interdependence: The fact that hA appears in Bridget’s production function and hB in Abdul’s represents the external effects and therefore the interdependence between the fishermen. The fact that the sign of these terms is negative means that the external effect is negative. We depict Abdul’s production function in the top panel of Figure 5.5. The higher of the two green curves represents the relationship between his labor input and his fish output when Bridget is not fishing at all, that is: hB = 0. (We will explain the lower curve in a moment.) Abdul’s production function is increasing but becomes flatter the more time Abdul fishes. The slope of this curve is the marginal product of time fishing, indicating for each level of hA the increase in the amount of his catch that would result if he increased his fishing time a little. We also call this the marginal benefit of fishing time because it indicates how much he benefits if he fishes a little more (how much the larger catch from additional fishing time raises his utility). The second, lower, dark green curve in the top panel shows how Bridget’s fishing for 12 hours reduces the amount of fish Abdul will catch for each C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 235 level of Abdul’s fishing time. Like the top curve, it is rising: as Abdul spends more time fishing he catches more fish. But two things about it are important: • it is lower than when Bridget does not fish, and • its slope is also lower (it is flatter for each hour that Abdul spends fishing). Both are the result of the negative external effect that Bridget’s fishing inflicts on Abdul. In the lower panel of Figure 5.5 we show the marginal product of an hour of fishing based on the production function shown in the top panel, labeled the marginal benefit of hours of fishing. In Figure 5.5 when Abdul fishes 12 hours a week (and Bridget does not fish), his catch is 288 but when she also fishes 12 hours (the lower green curve) his catch is just 216 lbs. Equally important, when Bridget is not fishing, and Abdul is fishing 12 hours, his marginal product is 18. The fact that the marginal benefit curve shifts downward when Bridget fishes 12 hours reflects the fact that in the top figure for any given amount of fishing time by Abdul, the curve is flatter. M-Note 5.2: Numerical Examples for Productivity and External Effects Throughout the chapter, we’ll cover a worked example where Abdul and Bridget have the same level of productivity and external effect on each other. We shall assume that a = 30 and that b = 12 . Abdul and Bridget’s utility functions therefore become the following: Abdul’s utility: uA (hA , hB ) = hA (30 Bridget’s utility: uB (hA , hB ) = hB (30 1 B (h + hA )) 2 1 A (h + hB )) 2 1 A 2 (h ) 2 1 B 2 (h ) 2 (5.11) (5.12) In the case where the fishermen fished alone, that is the other fishermen had zero hours fishing, Abdul’s utility would therefore be: uA = 30hA 12 (hA )2 12 (hA )2 = 30 (hA )2 . When Abdul and Bridget both spend time fishing, the external effect reduces Abdul’s utility, therefore he would have uA 30hA 1 A B 2h h = 30hA 1 A B 2h h 1 A 2 2 (h ) 1 A 2 2 (h ) = (hA )2 . 5.4 A best response: Another constrained optimization problem To understand the Nash equilibrium of the interaction between Abdul and Bridget we will need to know how each will best respond to any of the possible levels of fishing chosen by the other. This is because a Nash equilibrium is a mutual best response. To do this we will derive the best response function of each. But to do this we begin, as we did in Chapter 4, with a simpler problem: R E M I N D E R A player’s best-response function gives, for every possible set of strategy chosen other players player, the strategy that maximizes the player’s utility. A strategy profile in which all players are playing a best response, is a Nash equilibrium. 236 MICROECONOMICS - DRAFT showing how one of them, Abdul, will choose how many hours to fish, when Bridget is fishing at some given number of hours. Abdul, choosing a level of fishing time As a first step we bring together the information we have from the previous two sections on their preferences and their technology in a single equation expressing the benefits and costs of fishing: Utility = Total benefit (fish caught and consumed) Total cost (disutility of fishing time) So, for each of them, we substitute the production functions (Equations 5.7 and 5.8) for yA = xA and yB = xB into their utility functions for their total consumption (Equations 5.1 and 5.2). Doing this we obtain for each of them a single function showing how their utility depends on their own and the other’s fishing times, which is why we write their utility functions as uA (hA , hB ) and uB (hA , hB ) Abdul’s utility: uA (hA , hB ) = hA (a b (hA + hB )) Bridget’s utility: uB (hA , hB ) = hB (a b (hA + hB )) 1 A 2 (h ) 2 1 B 2 (h ) 2 (5.13) (5.14) For concreteness let’s suppose that Bridget is not fishing at all: she is a farmer and does not interact with Abdul in any way. We can therefore substitute hB = 0 into Abdul’s utility function, Equation 5.13. Then Abdul’s constrained optimization problem is to maximize his utility subject to the constraint given by how productive his fishing time is when Bridget is not fishing. This problem is set out in Figure 5.6 which combines Abdul’s indifference curves from Figure 5.4 with his production function (when hB = 0) from 5.5. Abdul might first consider fishing six hours, with results indicated by points f, g, and h in Figure 5.6. To determine if he should fish 6 hours he would compare: • the marginal cost of working more: namely the marginal disutility of working time, which is the slope of the indifference curve at f shown as point h in the lower panel with • marginal benefit of working more: namely, the marginal productivity of his fishing time, which is the slope of the production function at f shown as point g in the lower panel. From either the two slopes at point f (mrs 6= mrt ) in the top panel or their representation by points g and h in the bottom panel (mb > mc) Abdul would see he would increase his utility by working more than 6 hours. How much more? He will adopt the following method. He will compare: C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 237 Consumption, yA (pounds, lb) Figure 5.6: Abdul maximizes his utility subject to the constraint of his production function when Bridget does not fish. s 300 uA3 225 uA2 144 Feasible set, hB = 0 f uA1 0 A's marginal costs & benefits (lb) A's production function xA(hA, hB = 0) defines feasible consumption, yA 3 6 9 12 15 18 21 24 27 30 Marginal cost or disutility mcA = hA α g 24 s 15 mb = mc Marginal benefit when, hB = 0 6 h 0 3 6 9 12 15 18 21 24 27 30 A's hours, hA Marginal benefit (mb): The additional fish he would catch is the marginal benefit of fishing more. Marginal cost (mc): The additional effort he exerts or disutility it costs him is the marginal cost of fishing more. Abdul will best respond if he follows some simple rules: mb > mc If the marginal benefit exceeds the marginal cost as at point f, then fish more. mb < mc If the marginal cost exceeds the marginal benefit, then fish less. mb = mc If the marginal cost equals the marginal benefit, do not change how 238 MICROECONOMICS - DRAFT much you fish. 5.5 A best-response function: Interdependence recognized You can confirm from the figure that following the rule in italics just above, Abdul will fish 15 hours if Bridget is not fishing, indicated by point s in the figure (s for "solo" because Bridget is not fishing). This gives us just one point on his best response function hA (hB = 0) = 15 hours. What about when Bridget is fishing, for example, fishing 12 hours? Abdul’s reasoning is identical to the above rule. This case is illustrated in Figure 5.7 where the new feasible set constraining Abdul is smaller, because his catch for any amount of time that he spends fishing is reduced by Bridget also fishing. Abdul knows that the level of fishing that will maximize his utility under these new conditions is that which equates: • the slopes of an indifference curve and his production function so that the two are tangent in the top panel • or, to put it another way, the marginal benefit and the marginal cost of more fishing in the bottom panel. This gives us a second point on Abdul’s best response function, hA (hB = 12) = 12. Abdul fishes less when Bridget fishes more. This occurs because Bridget’s fishing more reduces the marginal benefit to Abdul’s fishing. What about Abdul’s response to Bridget fishing different hours. We do not have to go through the above process, tediously making a separate figure for each level of fishing time she might choose. Instead we can use mathematical expressions for the marginal costs and benefits of fishing to determine Abdul’s best response not as a discrete point, but as a continuous function, giving us his fishing time for any level of fishing Bridget might do. Using the rule that the best response is the number of hours that equates marginal benefits to marginal costs we have a general rule that can be expressed mathematically and which allows us to isolate hA as a function of hB and the parameters a and b . Here is the rule: a best response is a value of hA that satisfies the following rule: Marginal benefit a A B b (2h + h ) = Marginal costs = hA (5.15) Re-arranging Equation 5.36 to isolate hA and to express his optimal fishing C O O R D I N AT I O N F A I L U R E S Consumption, yA (pounds, lb) xA(hA, hB = 0) 300 uA3 225 uA2 144 uA1 A's marginal costs & benefits (lb) 0 uA3 uA2 s uA1 xA(hA, hB = 12) n Feasible set, hB = 12 3 6 9 12 15 18 21 24 Marginal benefit when, hB = 0 30 Marginal disutility mcA = hA 24 Marginal benefit when, hB = 12 15 s n 12 0 3 6 9 12 15 18 21 24 A A's hours, h hours as a function of Bridget’s hours hA (hB ), we have Abdul’s best-response function: hA (hB ) = a b hB 1 + 2b (5.16) How does Abdul’s fishing time hA change when the variable (hB ) and parameters (a and b ) change? • Change in Bridget’s fishing time (hB ): If Bridget decreases her fishing time, Abdul’s marginal benefit curve shifts up, and Abdul’s best response is to increase his fishing time to balance his marginal cost with the higher marginal benefit. Abdul’s best-response function does not shift, he chooses a different level of fishing due to the change in Bridget’s fishing time. & INSTITUTIONAL RESPONSES 239 Figure 5.7: Abdul maximizes his utility subject to the constraint of the production function when Bridget spends 12 hours fishing. The feasible set is now smaller because of the negative external effect that her fishing imposes on Abdul. In the top panel, at point n his indifference curve labeled uA1 is tangent to his production function, meaning in the lower panel, that the marginal disutility of fishing time is equal to the marginal productivity of fishing time, or the marginal cost of fishing more is equal to the marginal benefit. 240 MICROECONOMICS - DRAFT • Change in maximum productivity (a ): If Abdul’s basic productivity increases, and nothing else changes, this shifts his marginal benefit curve up and independently of any change in Bridget’s fishing time, he will increase his fishing time to balance his marginal cost with the higher marginal benefit. This is a shift in Abdul’s best-response function itself, not just a movement from one point on it to another as in the bullet above. • Change in the over-fishing effect (b ): If the external effect increases, Abdul’s marginal benefit curve pivots downward with a corresponding decrease in fishing time (b changes the slope of his marginal benefit curve, as can be seen from Equation 5.36). Like the increase in a , in this case Abdul changes his fishing time due to a shift in this best-response function. The best-response function for Bridget can be derived in the same way we derived Abdul’s. Therefore her best-response function is: Bridget’s BRF : hB (hA ) = a b hA 1 + 2b (5.17) M-Note 5.3: Marginal benefits, marginal costs, and finding the best responses In M-Note 5.2, we used the example of a = 30 and b = 12 to provide utility functions for Abdul and Bridget, as represented in equations 5.11 and 5.12. We now use those parameters to identify the first-order condition for Abdul’s utility maximization where his marginal benefits equal his marginal costs and therefore to provide a best-response function. uA (hA , hB ) = hA (30 ∂ uA ∂ hA = (30 | uAhA = 1 B (h + hA )) 2 1 B h 2{z hA ) } Marginal benefit 1 A 2 (h ) 2 hA |{z} =0 Marginal cost We can isolate Abdul’s hours of work, hA to find his best response to Bridget’s hours of work: Abdul’s BRF: Bridget’s BRF: hA (hB ) hB (hA ) = = 30 2 30 2 1 B 2h 1 A 2h = 15 = 15 1 B h 4 1 A h 4 (5.18) (5.19) Each of them therefore has a best-response function that is a function of the other person’s time spent fishing: hA (hB ) for Abdul and hB (hA ) for Bridget. M-Note 5.4: Mathematics of the best-response function To understand each player’s response to the other, it is useful to understand their marginal utilities of hours of fishing. We do this for Abdul, in the understanding that the Bridget will have symmetrical results. We will therefore find uA , Abdul’s marginal utility of his own hA hours of fishing, uA , marginal utility of Bridget’s hours of fishing, and hA (hB ), Abdul’s hB C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 241 best-response to Bridget’s choice of hours. We start with Abdul’s utility function: uA (hA , hB ) = hA (a b (hA + hB )) 1 A 2 (h ) 2 We can differentiate Abdul’s utility function with respect to his own hours (hA ) to find his marginal utility of his own hours of work. We also differentiate his utility function with respect to Bridget’s hours of work to find how his utility changes when Bridget changes her hours (hB ). A’s marginal utility of hA uAhA = Marginal effect on A’s utility of hB uAhB = A If we set Abdul’s marginal utility uA = ∂∂ uhA hA Bridget’s hours of work: ∂ uA ∂ hB = a b hB 2b hA = a b hB hA (1 + 2b )(5.21) = = a b hB = a b hB hA = a b hB (1 + 2b ) A s BRF: hA (5.20) b hA (5.22) = 0, then we can find his best response to ∂ uA ∂ hA hA (1 + 2b ) uAhA = Isolate hA term ∂ uA ∂ hA hA (1 + 2b ) = 0 Which is what we found from setting marginal benefit equal to marginal cost to find Abdul’s best-response function in Equation 5.16. We can use the above to define Abdul’s marginal rate of substitution: mrsA (hA , hB ) = = uAhA uAhB a b hB (1 + 2b )hA b hA (5.23) Checkpoint 5.3: How the BRFs change Consider how the parameters would change the best response functions in M-Note 5.2. a. What would happen if a = 30 changed to a = 24 and b = 12 changed to b = 13 ? b. What would the vertical and horizontal intercepts be for each of Abdul and Bridget? Sketch the best response functions you found in a. 5.6 How will the game be played? A symmetric Nash equilibrium We do not have enough information to answer the question in the section title. To do this we need answers to other questions. Is one of them powerful enough determine the allocation unilaterally, stating: I fish 15 hours, and you are excluded from fishing? Is there a government that can place a tax on fishing to discourage over-harvesting the stock? Can Abdul and Bridget agree to fish less? If they did, can they count on their agreement being enforced? In other words, we need to know more about the rules of the game. R E M I N D E R We began our analysis of Ayanda and Biko trading data and coffee in a similar way, with the two being symmetrical traders with neither of them having any particular advantage in the bargaining process. 242 MICROECONOMICS - DRAFT One possibility is that the two are independent (they do not make agreements with each other), self-regarding, and symmetrical, in that neither has any particular advantage in their interaction. So they simply try to do the best that they can, given what the other is doing and given the information they have. We will investigate other rules of the game later. A stationary allocation among symmetric players To study this case, we graph the two best-response functions in Figure 5.8. This gives us all the information we need to determine the Nash equilibrium of their interact on. A Nash equilibrium is a mutual best response, so Abdul’s choice of fishing hours must be a best response to Bridget’s choice of fishing hours, which must in turn be a best response to Abdul’s choice of fishing hours. This sounds complicated but with a little help from the mathematics we have already done, it is not: A Nash equilibrium is a point that is on both of the players’ best-response functions. We label the point n and define the hours that they work at the Nash equilibrium as (hAN , hBN ), where point n is the Nash equilibrium in the figure and the superscript N indicates each player’s Nash equilibrium hours. A Nash equilibrium is a pair of fishing times (hAN , hBN ) that satisfy each fisherman’s best-response function. When both players act according to their best-response functions, the outcome is a Nash equilibrium. In Figure 5.8 we plot the two best-response functions. The Nash equilibrium is the point where the two best response functions intersect the only point that the two lines have in common, shown by hAN and hBN . We show in the M-Note 5.5 how to find the Nash equilibrium hours of fishing for each person. At the Nash equilibrium, the two fishermen will spend the same amount of time fishing. hAN = hBN = a 1 + 3b (5.24) Equation 5.24 shows that each fisherman’s hours spent fishing is defined by the parameters a and b , capturing the effects on their best-response of their average productivity, their decreasing marginal productivity, and the negative external effect each has on the other. The Nash equilibrium fishing hours, hAN and hBN , are equal because the Abdul and Bridget have identical utility functions (other than reversing the superscripts), and they are determined by the parameters a and b . The greater is the maximum average productivity, a , the greater will be their equilibrium hours of fishing. The larger is the negative external effect each has on their own productivity and on the other person’s productivity, b , the lower their equilibrium hours will be. C O O R D I N AT I O N F A I L U R E S B's hours, hB α = 15 1 + 2β h 243 Figure 5.8: Nash equilibrium: mutual best responses for Bridget and Abdul. The equations for the best-response functions are: 24 BN & INSTITUTIONAL RESPONSES B's best−response function ● = 12 n hAN = 12 0 α = 15 1 + 2β 24 A's hours, hA M-Note 5.5: Finding Nash Equilibrium fishing time By definition of the Nash equilibrium, Abdul’s Nash equilibrium fishing time must be a best response to Bridget’s Nash equilibrium fishing time, and Bridget’s Nash equilibrium fishing time must be a best-response to Abdul’s Nash equilibrium fishing time. A Nash equilibrium is therefore a pair of fishing times (hAN , hBN ) that satisfy the following equations: hAN = hA (hBN ) = (a b hBN ) (1 + 2b ) (5.25) hBN = hB (hAN ) = (a b hAN ) (1 + 2b ) (5.26) Equations 5.25 and are two linear equations in two unknowns. We can solve the equations for the unknowns, which are the fishing times at the Nash equilibrium. There is a particularly simple way to do this in our case because: 1. The two fishermen have identical utility functions (they are mirror images of each other); so 2. we know that it must be that hAN = hBN , and 3. we can therefore set the Nash equilibrium level of fishing of the one equal to the best-response function of the other. So substituting hB = hA , into Abdul’s best-response function is: hA (hB ) Multiplying out and isolating hA : = a b hA 1 + 2b a b hA 1 + 2b hA hB = a b hB 1 + 2b If a = 30 and b = 0.5, the parameters we used in the previous figures, then we can see that when Bridget does not fish (the intercept of Abdul’s best response function with the horizontal axis) he fishes 15 hours. The point at which their bestresponse functions intersect is the Nash equilibrium of the interaction. Using these same parameters, we can see that the Nash equilibrium given by Equation 5.24 is that they both fish 12 hours. Nash Equilibrium A's best−response function 0 hB hA = 244 MICROECONOMICS - DRAFT hA + 2b hA = a A h + 3b h = a hA (1 + 3b ) = a AN = A h b hA a = hBN 1 + 3b 5.7 How would the players get to the Nash equilibrium? A dynamic analysis When we used the equation for the Nash equilibrium level of hours of fishing (Equation 5.24) to say what the effect of a change in a or b would be, we used what is called comparative static analysis. C OMPARATIVE STATICSWhen using comparative statics we compare the status quo outcome or the Nash equilibrium before the change with the outcome or Nash equilibrium after the change. When using comparative statics we compare the status quo outcome or the Nash equilibrium before the change with the outcome or Nash equilibrium after the change. • The word static refers to the Nash equilibrium because at a Nash equilibrium there are no reasons for the actors to change what they are doing. • The process is comparative because we compare two or more states before and after a change. We did the following: • We started by assuming that Aram and Bina were at the Nash equilibrium, each fishing 12 hours. • We then assumed that other things (like the weather) that might affect their fishing time are held constant (this is called the ceteris paribus assumption or ’other things equal’). • Then we compared the two Nash equilibria, one before the change and the other after the change, and. • We assumed that after the change Aram and Bina would be at the new Nash equilibrium, working some different number of work hours. • Finally we considered the difference in work hours between the two Nash equilibria to be the effect of the change on work hours. This type of analysis gets the name comparative static because it compares two static (unchanging) situations without looking at how the change takes place. This is an essential method of economic analysis, simplifying the matter by a shortcut. The shortcut is that we did not explore the process by which the move from the first to the second Nash equilibrium occurs, namely who did what to implement the move. This is called a dynamic analysis because it is R E M I N D E R When we say "other things equal" we are using the ceteris paribus assumption which allows us to compare what happens when one variable of interest changes. C O O R D I N AT I O N F A I L U R E S Marginal benefit Marginal disutility when, hB = 12 mcA = hA 24 mb > mc A mbA < mcA mbB < mcB mbA > mcA at 6 hours 12 mbA < mcA at 18 hours n 245 A's best−response function A B's hours, hB A's marginal costs & benefits (lb) 24 & INSTITUTIONAL RESPONSES mcB < mcB 15 n 12 B's best−response function mbB > mcB mbB > mcB mbA > mcA mbA < mcA 0 0 3 6 9 12 15 18 21 24 0 6 A's hours, hA (a) Marginal benefits and costs when B fishes 12 hours 12 15 18 24 A's hours, hA (b) Dynamic analysis of A and B’s choices based on the process of change. (The term dynamic refers to change, it is the opposite of static.) We did not even explain why Abdul and Bridget would have been at the original Nash equilibrium in the first place. Fortunately, the way we have derived our best responses provides a way to fill in the necessary dynamic analysis. Remember, when Abdul was selecting a best response he adopted a sim- Figure 5.9: How players can get to the Nash equilibrium: A dynamic analysis. Panel a. shows the marginal costs and benefits of Abdul’s fishing if Bridget fishes 12 hours. In panel b. shows the dynamics of the choices in terms of the fishermen’s marginal benefits and marginal costs. The horizontal arrows show the direction Abdul will move if he is initially at the base of the arrow. The vertical arrows show the same for Bridget. The inequalities involving marginal benefits and costs (mb, mc) are the reason for the movement shown in the arrows (which are called "vectors"). ple check list based on the marginal benefits of fishing more (mb) and the marginal costs of fishing more (mc): • if mb > mc, then fish more • if mb < mc, then fish less • if mc = mc don’t change how much you are fishing. When we introduced this check list we focused on the last line, because that is the equality that determines the utility-maximizing level of fishing for Abdul, that is, it is a point on his best-response function. The top two lines of the checklist tell Abdul what to change when he is not fishing the optimal amount given what Bridget is doing, that is when he is ‘off’ his best-response function. As Figure 5.9 shows, these first two lines of Abdul’s checklist tell us that starting at any allocation (that is any combination of fishing hours of each of them) in which direction he should move, shown by the arrows. The dynamic analysis gives the following simple instruction: if you are not on your bestresponse function, move toward it. Abdul’s arrows are green and horizontal (when he changes his fishing hours M - C H E C K Abdul might adopt the instruction: close half of the difference between the hours I am now working and the hours indicated by my best-response function, given how many hours Bridget is now working. For example, if Abdul were fishing 6 hours while Bridget fished 12 hours, he would increase his hours by (12 6)/2 = 3 hours. 246 MICROECONOMICS - DRAFT he moves left or right). The same reasoning allows us to show the dynamic arrows for Bridget, they are blue and horizontal, because when she changes her hours that moves the allocation point up or down. For example, in Figure 5.9 a, if Abdul is fishing 6 hours the marginal benefits of fishing more exceed the costs (the bracketed term on the left). So in Figure 5.9 b, the arrows show that he will fish more. Similar reasoning (in reverse) applies to the case where he is fishing, for example, 18 hours. The extent by which the benefits differ from the costs depend on how much fishing Bridget is doing. Figure a shows the case for when she is fishing 12 hours. You can also work out how Bridget will adjust her hours if she is fishing more or less than the amount indicated by her best response function. You can see from the figure that unless the allocation is at point n one or both of them will have an incentive to move (horizontally for Abdul, vertically for Bridget) in ways that will lead them to the Nash equilibrium. This explains why we would expect both Bridget and Abdul to be at (or very close to) the Nash equilibrium. It also explains, if the Nash equilibrium shifted because of some change in either a or b , why we would expect the two to alter their fishing hours to move towards the new Nash equilibrium. We now introduce a way that we can evaluate all of the possible equilibria of this game by the standards of Pareto efficiency and fairness. M-Note 5.6: Numerical Nash Equilibrium In M-Note 5.3, we found the best responses for Abdul and Bridget given by Equations 5.18 and 5.19. Using the method we outlined above, we set Abdul’s Nash equilibrium hours of fishing equal to Bridget’s BRF to find the Nash equilibrium level of fishing time: Bridget’s BRF : Collect terms Multiply by hAN 1 hA + hA 4 ✓ ◆ 5 A h 4 4 5 = 15 1 AN h 4 = 15 hAN = = 15 ✓ ◆ 4 15 = 12 = hBN 5 (5.27) As a result, we see that each will fish 12 hours at the Nash equilibrium. Therefore they each obtain the following Nash-equilibrium utility (by substituting hAN and hBN into their utility functions): uAN (hAN , hBN ) = = = 1 BN 1 AN 2 (h + hAN ) (h ) 2 2 1 1 12(30 (12 + 12)) (12)2 2 2 216 72 = 144 = uBN hAN (30 Each of them has a utility of 144 at the Nash equilibrium and the total welfare (sum of utilities) is W N = uAN + uBN = 288. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES Checkpoint 5.4: Storms and sustainability Imagine that the external effect increased, as it would, for example, if greater climate volatility produced storms that caused the two fishers fish in the same limited part of the lake. • Use the equation for the best responses of the two to redraw the figure. Why do the fishermen best respond by fishing less? • Use the equation for hAN and hBN to show that the Nash equilibrium level of fishing will decline. • Use what you have learned to explain how the best-response functions and the Nash equilibrium would change if the fishermen jointly adopt a strategy to let go of young fish to make the fish population more sustainable and reduce the external effect they have on the other fisherman. 5.8 Evaluating outcomes: Participation constraints, Pareto improvements and Pareto-efficiency Because the symmetrical interaction is just one of many possible rules of the game that Bridget and Abdul might engage in, we need to go beyond the Nash equilibrium for that game and think find a way to evaluate all of the possible allocations that they might experience. To do this, as in Chapter 4 we use the indifference maps of the two players superimposed on the same set of outcomes. Recall that in the previous chapter every point in the Edgeworth box indicated an allocation composed of a bundle of goods for Ayanda and another bundle of goods for Biko. We will see that the same is true in this case if we plot an allocation as the pair of fishing hours of the two, hA , hB . We start with Abdul’s preferences. Because the utility of each depends on their own fishing time and the fishing time of the other, that is because Abdul’s utility: Bridget’s utility: uA (hA , hB ) B A B u (h , h ) (5.28) (5.29) we can plot indifference curves with fishing time on each axis: Bridget’s fishing time (hB ) on the vertical axis and Abdul’s fishing time (hA ) on the horizontal axis. We do this in panel a of Figure 5.10, where every point in the figure is a particular allocation of fishing times (hA , hB ). Using these allocations we can calculate the utility that Abdul would experience were that allocation to occur. On this basis we can calculate Abdul’s indifference curves based on in his hours of fishing and Bridget’s hours of fishing. Abdul prefers curves labeled with higher numbers, u3 > u2 > u1 . Notice two things about the indifference curves: 247 248 MICROECONOMICS - DRAFT • The vertical dimension, or the effect of Bridget fishing more. Abdul’s preferred indifference curves are lower. This is because the less Bridget fishes the better it is for Abdul. • The horizontal dimension, or the effect of Abdul fishing more. If Bridget is fishing at the "low" level indicated in the figure and supposing that Abdul initially does not fish at all but considers fishing a little, he will start by finding himself at successively higher indifference curves as he fishes A more, crossing the indifference curves labeled uA 1 , and then u2 and up to uA3 . But if he spends too much time fishing he will then cross from uA3 back A down to indifference curve uA 2 and again go back down to u1 . Another perspective on a best-response function We can use the horizontal dimension of the figure to identify a point on Abdul’s best-response function, associated with Bridget hypothetically fishing just 8 hours. We take this thought experiment as a constraint on Abdul’s utility maximization. In the figure the green shaded area is his feasible set, and the horizontal line is its frontier. Remember Abdul prefers indifference curves that are lower down (indicating Bridget fishing less). The most preferred indifference curve that is feasible is the one tangent to the constraint, at point j, which is therefore a point in Abdul’s best-response function. It may help to think of his indifference map as showing the contours of the shoulder of a hill, and Abdul as walking along the frontier of the feasible set towards point j trying out different amounts of time he might devote to fishing. This is exactly what he did in Figure 5.6, comparing the marginal benefit and marginal cost of fishing more. At first he is climbing – crossing contours indicating ever-higher altitudes - higher utility. When he fishes 6 hours, he A achieves utility uA k = 121, proceeding on to fish 8 hours, he achieves un = 144 , and finally fishing 13 hours, he achieves uAj = 169. At point j his path levels off and if he continues to increase his fishing time he will descend to lower altitudes – lower utility – once more. Panel b of Figure 5.10 illustrates how two additional points on Abdul’s best response-function are derived. The best-response function is constructed by considering all of the possible levels of fishing that Bridget could hypothetically do, and then reason as we did for point j. Notice that Abdul’s best-response function intersects the indifference curves where the indifference curves are flat. If the indifference curve is flat then the mrs must be zero. In M-Note ?? we show why this must be true. In the right panel you can see that as Bridget’s fishing time increases from 8 to 16 hours, Abdul’s fishing time declines from 13 to 11 hours. He identifies his bestresponse hours of fishing by finding the point on his best-response function that corresponds to the number of hours Bridget fishes. C O O R D I N AT I O N F A I L U R E S k uAk = 121 12 n uAn 8 = 144 j uAj = 169 B's hours,hB B's hours,hB 16 hBHigh & INSTITUTIONAL RESPONSES A's best−response function = 16 k hBN = 12 n hBLow = 8 j uAk 6 8 13 0 the indifference curves are flat? The negative of the slope of the indifference curve is: = uAhA uAyA (5.30) Abdul’s best-response function gives the values of hA and hB for which the derivative of Abdul’s utility with respect to his fishing time is equal to zero or: If uA hA = a b hB 15 18 (b) Abdul’s best-response function M-Note 5.7: Why is the best-response function made up of points where = uAj A's hours, hA (a) Abdul making a choice constrained by Bridget’s hours uAhA uAn 11 12 13 A's hours, hA mrsA (hA , yA ) 249 (1 + 2b )hA = 0 0, then the numerator of Equation 5.30 is zero, so the slope of the indifference curve is equal to zero, which means that it is flat. Checkpoint 5.5: The marginal rate of substitution Although each of the fishermen doesn’t "control" the hours of work that the other does, we can still think in sensible ways about the marginal rate of substitution. a. Consulting Figure ??, what is the sign of the marginal rate of substitution going from left to right along the indifference curve leading up to point n on uA2 (on the left-hand side of point n)? Why? b. Continuing to consult Figure ??, what is the sign of the marginal rate of substitution going from left to right along the indifference curve on the right-hand side of point n on uA 2 ? Why? Fallback positions and the Pareto improving lens We have said that rules of the game other than symmetrical interaction will lead to different Nash equilibrium allocations. As long as the interaction among the two is voluntary – there are no "offers you cannot refuse" – we Figure 5.10: A new look at Abdul’s constrained optimization problem for selecting his fishing time depending on Bridget’s fishing time. . To illustrate the construction of Abdul’s bestresponse function, in Panel a we consider Abdul’s decision about how many hours to work, given that Bridget has (hypothetically) decided to work 8 hours. The horizontal blue line is the constraint on Abdul’s utility-maximizing process. In panel b, we consider three hypothetical levels of Bridget’s fishing time. The horizontal lines represent Bridget’s fishing time at each of these levels, and are the constraint on Abdul’s maximization process. One of these horizontal lines is tangent to each of Abdul’s indifference curves hB = 8 tangent to uAj at point j, hB = 16 tangent to uAk at point k, and hBN = 12 tangent to uAn at point n. Abdul’s entire best response function is made of of points like j, n, and k, for each of Bridget’s possible levels of fishing hours. 250 MICROECONOMICS - DRAFT can limit the possible outcomes by thinking about the alternatives that the two have should they decide not to fish at all. Any allocation in which they both fish and that makes either of them (or both) worse off than how they would do if they did not fish at all will not occur for the simple reason that they will not fish if they could do better by not. We have already introduced the idea that Abdul, if he does not fish at all, will receive an income of yz possibly from family, friends or the government. Suppose the same opportunity applies to Bridget. If they do not fish and receive yz then their utility is just yz . This is their fallback position (like the allocation z in the Edgeworth box of the previous chapter). But they only receive their fallback if they do not fish, so the opportunity cost of fishing – what they cannot have if they fish – is yz . In Figure 5.11 we show both Abdul’s and Bridget’s indifference maps. We see from the numbering of the utility labels on the curves, that Bridget’s indifference curves give greater values the closer they are to the vertical axis (as Abdul’s did with the horizontal axis). One of their indifference curves is particularly important: it is labeled uA z and uBz . These two curves show all of the allocations hA , hB that yield, for Abdul and Bridget respectively a level of utility equal to the utility of their fallback position namely uz = yz . This is the participation constraint for each of them: they will not participate in fishing unless they can do at least this well. Any point between these indifference curves is a Pareto-improvement over their fallback position: both are better off than their fallback option. Any point between these indifference curves is a Pareto-improvement. The Pareto-improvements are shown by the Pareto-improving lens shaded in yellow. The Pareto-efficient curve There is another important curve in Figure 5.11: the purple solid and dashed Pareto-efficient curve. We know that Pareto-efficiency requires that the fishermen’s indifference curves be tangent, that is, for their marginal rates of substitution to equal. You can see two of these tangencies in the interior of the Pareto improving lens. The other tangencies defining the Pareto-efficient curve are not shown. The Pareto-efficient curve is made up of all points representing allocations for which: mrsA = uAhB uAhA = uBhB uBhA = mrsB (5.31) Equation 5.31 shows the condition that the marginal rates of substitution must be equal at Pareto-efficient outcomes, meaning that their indifference curves are tangent. C O O R D I N AT I O N F A I L U R E S B's PC 21 uB3 = 155.8 B's hours, hB 18 uBz = 112 Pareto−efficient curve 15 A's PC 12 9 uAz = 112 6 3 uA3 = 155.8 uA1 = 144 0 0 3 6 9 251 Figure 5.11: Abdul’s indifference curves and Bridget’s indifference curves showing their fallback levels of utility (their participation constraints), uAz and uBz . At their fallback positions, they are not fishing at all and receiving a payment (in fish) of uAz and uBz from the government . uB1 = 144 24 & INSTITUTIONAL RESPONSES 12 15 18 21 24 A's hours, hA The figure clarifies the difference between Pareto improvements and Pareto efficiency : • The points on the purple Pareto-efficient curve that are indicated by a dashed line outside the yellow Pareto improving lens are Pareto-efficient but not Pareto improvements over the fallback no fishing option. • The points in the yellow Pareto-improving lens that are not on the purple Pareto-efficient curve are Pareto improvements but not Pareto efficient. Checkpoint 5.6: Understanding the parameters a. What would be the effect on the Pareto-improving lens if a increased? b. What would be the effect on the Pareto-improving lens if b increased? c. Why does b affect the person fishing even when no one else fishes? Why does it make economic sense? 5.9 A Pareto inefficient Nash equilibrium We return now to the symmetrical interaction between Bridget and Abdul, in which the Nash equilibrium is the allocation at the intersection of their bestresponse functions. And we ask: is that allocation Pareto-efficient? To answer, we combine two figures we have already introduced: Figure 5.11 showing the two fishermen’s indifference curves and Figure 5.8 showing their 252 MICROECONOMICS - DRAFT best-response functions. The combination of these figures results in Figure 5.12. Figure 5.12 shows that the Nash Equilibrium is not Pareto-efficient: at the Nash allocation (point n) the indifference curves of the two intersect rather than being tangent. So allocation n cannot be Pareto-efficient. How do we know that their indifference curves cannot be tangent at that point that is, how do we know that mrsA = uAhB uAhA 6= uBhB uBhA = mrsB (5.32) The answer is that the Nash equilibrium is a point on both best-response functions, defined by uB = 0 for Bridget’s best response and uAhA = 0 for hB Abdul’s best response. At the best response each fisherman adjusts their own fishing time to maximize utility so that these two terms will be zero. If we substitute the zeroes for the marginal utilities in Equation 5.31, we find the following: • The first expression is now zero divided by uB so the slope of Abdul’s hA indifference curve is zero; it is flat (as in Figure 5.12), and • The second expression is now uA divided by zero, so the slope of Abdul’s hB indifference curve is infinite; it is vertical (as in Figure 5.12 too) A flat line cannot be tangent to a vertical line, so the condition for Pareto efficiency is violated and the Nash equilibrium is not Pareto efficient. A view from a Pareto inefficient status quo Nash equilibrium. We now imagine Abdul and Bridget, fishing 12 hours each as indicated by the Pareto-efficient Nash equilibrium. They realize they could both do better. And they consider the options. They each might propose some different allocation. To agree on an alternative level of fishing, the proposal would have to implement a Pareto improvement. The Pareto improvement would need to be over the Nash equilibrium, not over their no-fishing fallback option. Remember that the Nash equilibrium is already better than their the fallback positions. With allocation n the new fallback for the agreement, we now have a new yellow shaded Pareto-improving lens. There are two things to notice about Pareto-improvements over the Nash allocation: • both fishermen spend less time fishing and both are better off (have higher utility than at the Nash equilibrium) and • the new Pareto-improving lens is much smaller than the lens of Pareto improvements over the no-fishing fallback option. R E M I N D E R To understand Figure 5.12 it will help to remember that for Abdul down is better (his indifference curves have higher utility the lower they are) because the less Bridget fishes the better it is for him. Similarly, Bridget is better off on the indifference curves further to the left. C O O R D I N AT I O N F A I L U R E S uB3 uBn B's hours, hB Pareto−efficient curve ● hB* = 10 ● Figure 5.12: The Nash equilibrium and the Pareto-improving lens. The Pareto-improving fishing times (in which both fish less) are in the pale yellow lens. Notice that Abdul’s indifference curve at the Nash equilibrium is flat, and Bridget’s at the same point is vertical (their marginal rates of substitution are not equal). This being the case there must be a Pareto-improving lens and the Nash equilibrium cannot be Pareto-efficient. n i A's best−response function uAz uAn 253 uBz B's best−response 15 function hBN = 12 & INSTITUTIONAL RESPONSES uA3 hA* = 10 hAN = 12 15 A's hours, hA Checkpoint 5.7: Checking the marginal rate of substitution Use the values of a = 30 and b = 12 and substitute them into the marginal rate of substitution for Abdul and Bridget. a. Confirm that when Abdul sets his mb = mc, the numerator is zero. b. Also confirm that if you find the equivalent for Bridget, the mrsB = • (or it is undefined) when Bridget sets her marginal benefits equal to her marginal costs. The reason why there exist allocations that are Pareto improvements over the Nash is as follows. • Reason 1: Each of them would benefit a lot if the other were to fish less and • Reason 2: at the Nash equilibrium each of them would experience very little lost utility by themselves fishing a little less. Reason 1 concerns each fishermen’s marginal utility with respect to the other’s hours of fishing, and we can see that uA < 0 and uBhA < 0 because hB each fisherman’s fishing time reduces the other’s productivity. Concerning Reason 2, suppose that, at the Nash equilibrium level of the fishing times, Bridget decided she would try to bribe Abdul to fish less. How much would she have to give him to fish a tiny bit less? The answer is "almost nothing" because at the Nash equilibrium, changes in his fishing time have no 254 MICROECONOMICS - DRAFT effect on his utility because the marginal benefits of fishing a little more or less equal the marginal costs of fishing a little more or less (that is how he chose that level of fishing to do). So Abdul’s fishing a little less would not matter much to Abdul but it would definitely benefit Bridget. A similar results is true for Bridget: Abdul could bribe her to fish a little less for a tiny portion of his fish. This being the case if they both could agree to fish less (and just forget about the bribes) they would both be better off. The conclusion is that Bridget and Abdul need not lament their sorry condition at the Nash equilibrium. If a deal can be enforced – an agreement to limit fishing, maybe along with a bribe – there’s a deal to be made that benefits them both. We turn now to considering changes in the rules of the game that might reduce fishing times, keeping in mind that we are thinking about not just two people, but an entire community of people – perhaps the entire world’s population if we are considering coordination problems such as climate change or the spread of epidemic diseases. M-Note 5.8: The Nash equilibrium cannot be Pareto-efficient To show that the Nash equilibrium is not Pareto-efficient we ask if they could agree each to fish an arbitrarily small amount less would they both be better off. If the answer is "yes," then the NE cannot be Pareto-efficient. We know that uA < 0 and uBhA < 0 each would be hB better off if the other fished less. We also know that uA = 0 and uBhB = 0 because these hA equalities define Bridget’s and Abdul’s best-response functions, and the Nash Equilibrium they are trying to improve on is a pair of strategies each of which is a best response to the other. So for any change dhA and dhB , representing an agreement to change their fishing time, we can evaluate the change in each utility associated with change in the fishing times of each. duA = uAhA dhA + uAhB dhB duB = uBhA dhA + uBhB dhB Eliminating the terms equal to zero in the expressions above, namely those involving uA hA and uB we have: hB duA = uAhB dhB < 0 duB = uBhA dhA < 0 or, rearranging duA = uAhB < 0 dhB duB = uBhA < 0 dhA C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 255 Both expressions are negative: the utility of each would be enhanced by an agreement to fish a little less. The Nash equilibrium allocation of fishing times is not Pareto-efficient. Checkpoint 5.8: Pareto-improvements Use the numerical values of a = 30 and b = 12 a. How much utility would the players have if they both simultaneously reduced their hours of work from the Nash equilibrium values of 12 hours to 11 hours? Would they be better or worse off? b. If one reduced their hours to 11 hours, what would the other’s best response be? Would they actually reduce their hours to 11 or not? 5.10 A benchmark socially-optimal allocation To provide a benchmark or standard against which we might evaluate the various rules of the game that might improve on the Nash equilibrium of the symmetric interaction above, we will reintroduce the Impartial Spectator, who we relied on for the same purpose in Chapter 4. The Impartial Spectator wishes to determine fishing time and distribute fish so as to maximize a social welfare function, which, because she values the utilities of the two equally, is just the sum of the utilities of the two: Total social welfare W = Abdul’s utility + Bridget’s utility = uA (hA , hB ) + uB (hA , hB ) (5.33) She knows that the solution to this problem must be Pareto-efficient, because if it were not, then one of the two could be made better off without worsening the condition of the other, so this could not be the optimum for the Impartial Observer, who values the well being of both. This means that the socially optimal allocation must be somewhere along the Pareto-efficient curve in Figure 5.12. But where? A socially optimal allocation To answer the question, we transform the view of the problem in Figure 5.12, where the space in the figure is defined for hours of fishing, into a new graph, Figure 5.13, in which presents the same information in terms of the utilities of the two. The Pareto-efficient curve in Figure 5.12 appears in Figure 5.13 as the dark green curve that is frontier of the feasible set of utilities. Its slope is the marginal rate of transformation of Bridget’s utility into Abdul’s utility. This provides the answer to the question: along the feasible frontier, how much does Bridget’s utility have to fall in order for Abdul’s to increase by one unit? M - C H E C K In Chapter 4, we gave the Impartial Spectator Cobb-Douglas preferences over the two players’ utilities. Here we have the Impartial Spectator sum the utilities of the two fishermen, as the Impartial Spectator did in Chapter 3. 256 MICROECONOMICS - DRAFT W3 = 350 A's PC B's utility, uB 225 W2 = 300 W1 = 250 i 150 112 B's PC z Utillity possibilities frontier Feasible utilities 112 150 A's utility, uA 225 You can see that when Bridget has almost all of the feasible utility then it does not ‘cost’ Bridget much for Abdul to have a little more (the frontier is not very steep); but the marginal rate of transformation rises (the curve steepens) as Abdul gains more utility. The reason is that when Bridget has most of the utility, she is working long hours (almost 15) and incurring a substantial disutility of working time as a result. Fishing a little less would not reduce her utility much, but for Abdul fishing a little more would substantially increase his utility. So when Bridget is doing most of the fishing (and gaining most of the utility) the opportunity cost of increasing Abdul’s utility (in terms of Bridget’s forgone utility) is small. The Impartial Spectator’s values are expressed by her indifference curves (the blue lines), their slopes, her marginal rate of substitution, are a constant, namely 1, because she values the utility of the two equally. The point z represents the fallback utilities of the two (namely 112), and the yellow shaded area is the set of feasible Pareto-improvements over this fallback position. You can see that the optimum point, i, is found where the highest feasible Impartial Spectators indifference curve is tangent to the utility possibilities frontier (the frontier of the feasible set). So this is another case of the familiar mrs = mrt rule, but now for the Impartial Spectator, rather than Abdul or Bridget. marginal rate of substitution = marginal rate of transformation (5.34) Figure 5.13: Feasible utilities, the utility possibility frontier, and the Impartial Spectator’s iso-social welfare indifference curves. Here we show the utility possibilities frontier and feasible utilities for the Impartial Spectator. All points on the frontier are Pareto-efficient. The points above and to the right of the fishermen’s participation constraints constitute the bargaining set, that is the outcomes that are Pareto-superior to their fallback options, uAz = uBz = 112. The Impartial Spectator’s iso-social welfare indifference curves show her equal valuation of the utility of the two and the negative of the slope of her iso-social welfare curves is her marginal rate of substitution. The slope of her iso-social welfare curves is -1 indicating that she values the two utilities equally. The negative of the slope of the utility possibilities frontier is the marginal rate of transformation of Bridget utility into Abdul’s. That is, it is the opportunity cost of Abdul having more utility in terms of the utility that Bridget forgoes as a result. The impartial spectator will therefore choose point i where mrs = mrt to maximize social welfare given the constraint of the utility possibilities frontier. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 257 Rules that implement the social optimum We know that acting on the basis of their best-response functions Abdul and Bridget over-exploit the resource. They could both do better if they adopted a different rule for deciding how much to fish. Before turning to institutions that might implement such a new rule for their decisions, lets think about a rule that would exactly implement point i in the figure. To find the optimum, point i in Figure 5.13 the Impartial Spectator proposes the following rules that, if followed, will maximize her social welfare function Equation 5.33. We show in the M-Note 5.9 how these are derived. hA = a B h =a 2b hB 2b hA A B 2b h (5.35) 2b h Focusing on the equation for Abdul, the socially optimal condition looks very similar to Abdul’s best-response function (shown again below) when he was maximizing his own utility, except for one big difference. Marginal private costs Abdul’s own optimality condition h A = Marginal private benefits b 2hA = a b hB (5.36) Comparing Equations 5.35 and 5.36 with the latter rearranged, or: Marginal social costs Abdul’s social optimality condition A A h +bh we see that the difference is that there is an extra = Marginal private benefits = a 2b hB b hA (5.37) b hB in the socially optimal condition (Equation 5.35.) This is the negative external effect of Abdul’s fishing on Bridget’s utility. In Equation 5.37 we have moved this term to the left-hand side of the equation, adding it to the marginal private cost of fishing more (namely the disutility of hours of fishing, hA ). So Equation 5.37, the condition for Abdul’s fishing time to implement a social optimum says the following. marginal private cost + marginal external cost = marginal private benefit(5.38) The left-hand side is called the marginal social cost. The private cost (marginal or average) is the cost that the decision-maker bears as a result of some action that he or she takes. The social cost is the private cost plus any costs imposed on others as negative external effects. The Impartial Spectator reasons that if we are to implement an allocation that values the utility of both equally, then Abdul should act is if he is taking account of this cost – treating the costs he imposes on Bridget no differently than his own disutility of labor – when deciding on how much to fish. As you P RIVATE AND SOCIAL COST The private cost (marginal or average) is the cost that the decision-maker bears as a result of some action that he or she takes. The social cost is the private cost plus any costs imposed on others as negative external effects. 258 MICROECONOMICS - DRAFT know from the previous chapter, this is called internalizing the negative external effect of his actions. We can think of these socially optimal responses in the following way: Optimal response = Own best response + Internalized cost to other (5.39) Imposing the same condition on Bridget, the Impartial Spectator provides a rule where each fisherman internalizes the negative external effect of their hours of fishing on the other. As a result, we arrive at the levels of socially B optimal fishing time for Abdul and Bridget, denoted by as hA i and hi : Abdul’s socially optimal fishing time : hAi = Bridget’s socially optimal fishing time : hBi = a 1 + 4b a 1 + 4b (5.40) (5.41) Because b > 0, we see that each of the players’ Nash equilibrium levels of fishing time are higher than the socially opimal levels: hAN = hBN = a a > = hAi = hBi 1 + 3b 1 + 4b These socially optimal levels of fishing time correspond to point i (for impartial) in Figure 5.12 where Abdul and Bridget have the same fishing time (10 hours) and the same level of utility. The job description of the Impartial Spectator is not to figure out how this optimal allocation might be implemented ("way above my pay scale," she says). We leave that to a second (also imaginary) person who we will introduce in the next section. M-Note 5.9: The Impartial Spectator’s Choice We know that the Impartial Spectator has the following social welfare function, and we can substitute the fishermen’s utility functions into W as follows: Total Social Welfare = Abdul’s Utility + Bridget’s Utility W = uA + uB = hA (a = ahA + ahB b (hA + hB ) + hB (a 2b hA hB b (hA + hB )) 1 A 2 (h ) 2 1 B 2 (h ) 2 1 B 2 (h ) 2 1 A 2 (h ) 2 (5.42) Next, the social planner needs to find the social welfare maximum, or the optimal social welfare. To do this, we partially differentiate W with respect to the hours of fishing of each fisherman, hA and hB , as follows to find the first order conditions for the social welfare C O O R D I N AT I O N F A I L U R E S optimum: WhA = WhB = ∂W ∂ hA = hA = ∂W ∂ hB = hB = a 2b hB 2b hA a b hB b hB 1 + 2b a 2b hA a b hA b hA 1 + 2b hA = 0 (5.43) 2b hB hB = 0 (5.44) Notice that Equations 5.50 and 5.44 look similar to the best-response functions we found previously, but that each incorporates an additional b hB for Abdul in the numerator of Equation 5.50 and b hA for Bridget as shown in the numerator of Equation 5.44. These terms correspond to the cost of the external effect that each fisherman imposes on the other. M-Note 5.10: Numerical Choice of the Impartial Spectator The Impartial Spectator has a social welfare function, W which is the sum of the two fishermen’s utilities. We substitute each fisherman’s utility function (Equations 5.11 and 5.12) into the social welfare function and assume the parameter values of a = 30 and b = 12 . W = uA + uB = hA (30 = 1 B 1 A 2 (h + hA )) (h ) + hb (30 2 2 30hA + 30hB hA hB 2(hA )2 2(hB )2 1 A (h + hB )) 2 1 b 2 (h ) 2 (5.45) The Impartial Spectator then needs to differentiate the social welfare function defined by Equation 5.45 with respect to the two hours of work to determine how much each person should work: WhA = WhB = ∂W ∂ hA = ) hA = ∂W ∂ hB = ) hB = hB 30 hB 30 2 2hA = 15 hA 30 hA 30 2 1 B h 2 (5.46) 1 B h 2 (5.47) 2hB = 15 We can solve for the Impartial Spectator’s choice of work hours for Abdul and Bridget by substituting Equation 5.47 for hB into Equation 5.46. Following the process of substitution as usual, we find: hAi = 15 hAi = 15 = 7.5 3 A i 4 4 Multiply by hA 3 i hAi Collect terms = = 1 (15 2 1 A h ) 2 i 1 7.5 + hAi 4 4 30 (7.5) = 3 10 10 = hBi Each of them will work 10 hours at the socially optimal and Pareto-efficient allocation of B hours that the Impartial Spectator made. At the allocation (hA i , hi ), each of the fishermen has a utility of 150, which is higher than they had at the Nash equilibrium (see Marshal B Memo 5.6) and the total social welfare is Wi = uA i + ui = 150 + 150 = 300. & INSTITUTIONAL RESPONSES 259 260 MICROECONOMICS - DRAFT Checkpoint 5.9: Changing things for the impartial spectator Assume that a = 24 and b = 13 instead of the values you have used so far. 1. What would the number of hours be at the Nash equilibrium? How much utility would each player have and what would the total utility be? 2. How many hours would the Impartial Spectator choose? How much utility would each player have and what would the total utility be? 3. Sketch the best-response functions you found for (a) and sketch indifference curves for the Nash equilibrium and the Impartial Spectator’s choice through the relevant points. Remedies: Preferences, power, and policy Stories about two fictional people such as Abdul and Bridget and their textbook lake are light years away from real fishermen in Rhode Island or Australia. John Sorlien, the Rhode Island lobsterman who we quoted at the start of the chapter, is in competition not with a single other fisherman, but with hundreds. Unlike Abdul and Bridget (so far) real-world fishermen and lobstermen do sometimes cooperate to pursue common objectives (Sorlien headed their association). A friendly conversation and a handshake might be enough for Bridget and Abdul. But how might such an agreement be arrived at, and how might it be enforced in Rhode Island or Australia? An even greater challenge is how to design and enforce similar agreements – for example to burn less carbon in order to mitigate climate change – that span not thousands of actors but billions living under the jurisdiction of hundreds of independent governments. But the parable of Abdul and Bridget has provided an important insight. Illuminating the basic source of coordination failures: the negative effect of their own fishing on the other person (uA and uB , that is) is not part of the hB hA utility-maximizing process by which each choose how much to fish. Addressing these external effects is where institutions come in, meaning changes in the rules of the game. There are three basic approaches whether the common property resource that is being over exploited be a fish stocks in a lake or the limited carbon emissions carrying capacity of earth’s atmosphere. • Regulation of the exploitation of the resource by a government. • Private ownership of the resource so that private incentives will deter overexploitation. • Management of the resource through local interactions among the the H I S TO RY In 1968, Garrett Hardin wrote that “freedom in the commons means ruin to all" and as a result he advocated – “mutual coercion mutually agreed upon." But Hardin’s pessimism overlooked the many non-coercive ways that local communities have prevented the tragedy.5 C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 261 resource users. These three approaches are sometimes referred to as states (meaning governments), markets, and communities, or similar terms.6 5.11 Government policies: Regulation and taxation To underline the fact that we are not modeling actual governments but instead an ideal of what a well informed and set of public officials seeking to implement better allocations might do, we will introduce a second hypothetical character, the Mechanism Designer. He is tasked with finding policies to implement outcomes that would be recommended by the Impersonal Spectator according to her values of efficiency and fairness. He is an economist, with an engineering mentality, she is a philosopher and her job is to identify good, better or best outcomes. The mechanism designer’s job then is to implement the best that can be done. The Mechanism Designer is the main character in the Chapter 16 which is about public policy. You than think about him as an economist advising a government about how to design and implement its policies. A government has many options to address coordination failures such as common property resource over-exploitation, including educating the public about the costs and promoting basic research to find ways of making the resource more sustainable. But we focus on just two: • Fiat: Governments can sometimes order the implementation of an allocation – for example, reduced fishing – that they or the voters who elected them prefer. A fiat is an order. • Taxation: Governments regularly implement taxes as incentives: taxes make activities more costly and can therefore, discourage these activities, while still allowing each person or firm to choose how much of the activity to engage in given the increased costs. Fiat power With respect to fiat, the government, if it knew all the relevant information, B B could select hA = hA i and h = hi to maximize total utility. The government might implement this outcome by direct regulation, simply issuing a fishing permit allowing each fisherman a certain number of hours. Any deviation from the permitted hours would result in revocation of the permit and the fishermen would have to revert to their no-fishing fallback positions. Point i in Figure 5.12 is the Impartial Spectator’s efficient fishing time allocation that corresponds to what the government would choose. Assuming the government had no reason to favor one fisherman over the other from the F AC T C H E C K In response to the rapid depletion of fish stocks in 1992 the Canadian government simply banned fishing for cod in the Grand Banks region of the North Atlantic. Stocks have been recovering since then.7 262 MICROECONOMICS - DRAFT standpoint of fairness, a Pareto-efficient and equal distribution of fishing times would be the fiat allocation. Optimal taxes: Internalizing external effects Rather than implementing the efficient fishing time plan by fiat, however, the government might want to let the fishermen each decide how much to fish, but to change their incentives in order to address the coordination failure. The TAXES A tax is a charge the government enforces on the production or purchase of a good. A subsidy is a payment the government makes to the producer or purchaser, similar to a negative tax. government would levy what is called a Pigouvian tax on fishing designed to eliminate the discrepancy between the social and private marginal costs and benefits of fishing. The problem for the government is to select a tax rate on fishing time that as an intended byproduct will motivate the fishers to implement an allocation that maximizes total utility while at the same time maximizing their own utility. This means bringing the fishermen’s private incentives (the utility function that each maximizes) into alignment with the conditions laid out by the Impartial Spectator. The problem can be posed this way: find the tax rate that would transform the utility functions of the two fishermen so that their individual best-response functions are identical to those implied by the problem solved by the Impartial Spectator: maximizing total utility and internalizing the costs of the negative external effects. To internalize the cost means to require each of them to pay (in taxes) for the H I S TO RY Imposing taxes on particular behaviors which the government wants discourage because they impose negative external effects on others – over fishing fishing, smoking – the government is taking an approach pioneered by the early 20th century economists Alfred Marshall (1842– 1924) and A.C. Pigou (pee-GOO) (1877– 1959). In recognition of his contribution to the field of what is called welfare economics, these are sometimes called Pigouvian taxes. reduction in the catch of the other that their additional fishing time imposes. We know that each additional hour that Bridget fishes means that Abdul catches b hA pounds less of fish. So to force Bridget to take account of this negative external effect, she must be taxed at a rate of b hA for every hour she fishes. Bridget’s socially optimal tax rate depends on Abdul’s fishing time, because the external effect of Bridget’s fishing on Abdul’s well-being depends on how much Abdul fishes. If Abdul is not fishing at all, for example, there is no need to tax Bridget’s fishing, because it has no external effect. The tax that induces the fishermen to choose the socially optimal levels of fishing time is just equal to the negative external effect they impose on others at the Pareto-efficient levels of fishing time. A Pigouvian tax is a change in the rules of the game that has the effect of internalizing the external effect that is the cause of the coordination problem. The tax is an indirect form of coordination: the fishermen as citizens elect a government which they delegate to impose on them a set of incentives to overcome the over-fishing problem. E X A M P L E The "golden rule" is a common ethical principle that people should treat each other as they themselves would like to be treated. A Pigouvian tax is designed to accomplish the same result by imposing on each decision maker the costs that their decisions impose on others. C O O R D I N AT I O N F A I L U R E S M-Note 5.11: The best-response function of fishing with taxes Equation 5.48 shows Bridget’s utility function when her fishing time is taxed at the rate of t per hour fished: uB (hA , hB , t ) = hB (a b (hA + hB )) thB (hB )2 2 (5.48) To obtain the best-response function Bridget’s fishing conditional on the tax rate and Abdul’s fishing time, we differentiate the equation above and set the result equal to zero: uBhB uB hB = b hA a 2b hB hB = 0 t We can rearrange this first order condition to say that (on the left-hand side of the equation below) the marginal benefits of fishing more (in fish caught) must be equal to (on the right hand side) the marginal costs including the the disutility of additional fishing time plus the taxes incurred by fishing more: a b hA 2b hB = t + hB Re-arranging this to isolate hB so as to have a best-response function we have: (1 + 2b )hB = hB (hA , t ) = a a t b hA t b hA 1 + 2b (5.49) M-Note 5.12: A implementing the Impartial Spectator’s choice We know from M-Note 5.9 that the first order condition for Bridget’s fishing time in the social optimum allocation proposed by the Impartial Spectator is: hB = a b hA b hA 1 + 2b (5.50) The mechanism designer’s job is to find the tax rate per hour of Bridget’s fishing time that will induce her to act as if that were her private (self-regarding) first order condition too. We know from M-Note 5.11 that Bridget’s true best-response function including taking account of the tax is the following: hB (hA , t ) = a t b hA 1 + 2b (5.51) The question that the mechanism designer must now solve is: what is the level of the tax rate t , that will make Equation 5.51 look like Equation 5.50 so that Bridget’s private incentives will lead her to implement the Impartial Spectator’s social optimum. Comparing the two equations you can see that setting the tax rate that Bridget pays t B = b hA will make the two equations identical. So that is the optimal tax rate. The tax rate for Abdul would, by the same reasoning be t A = b hB . Then we can calculate the tax rate that Bridget pays at the Nash equilibrium. Because we know that the optimal tax implements the social optimum recommended by the Impartial Spectator, we substitute the value for hA i into the expression for the tax rate so we have , t = b hAi or Bridget’s tax rate Abdul’s hours worked (Nash) Bridget’s tax rate (Nash) t = hAi = thB = b hAi a 1 + 4b ab 1 + 4b & INSTITUTIONAL RESPONSES 263 264 MICROECONOMICS - DRAFT For the parameters we have introduced to illustrate the model with b = 0.5 and a = 30, the tax rate at the Nash equilibrium of the game with optimal taxes is 5 pounds of fish per hour of fishing time, meaning that every hour that Bridget fishes she must pay from her catch a total of five pounds of fish. Given that she is fishing 10 hours at the equilibrium of the game she pays 50 pounds of fish in taxes. Abdul pays the same amount. We do not subtract this amount from the their utilities at the Nash equilibrium because we assume that the total tax revenues are redistributed to the population equally without regard for their fishing times. Checkpoint 5.10: A partial tax Assume a = 30 and b = 12 . Consider that a government implements a tax where each person is charged t = 14 for an hour of work. a. What is Abdul’s utility function incorporating this tax? What is Bridget’s utility function with the tax? b. What are their best responses to the tax policy? c. What is the Nash equilibrium level of hours with t = 14 ? Is the outcome Pareto-efficient? If it is not Pareto-efficient, is it a Pareto-improvement over the Nash equilibrium without the tax? Explain. d. If t = 14 is insufficient to produce a Pareto-efficient outcome, what level of t would produce the Pareto-efficient outcome? Why? Explain. 5.12 Private ownership: Permits and employment But government policies are not the only change in the rules of the game that might address the over-fishing coordination problem. Suppose that the property rights over the lake are changed such that the lake is no longer a common pool resource, but is privately owned. As a result, lake is no longer non-excludable, as a resource that is both excludable and rival, the it is a private good. The person who owns the lake, say Bridget, could exclude Abdul entirely (remember that is what private property means). But as an owner she now has bargaining power over Abdul, and may be able to do better by letting him fish under conditions favorable to her. How do these new rules of the game change the Nash equilibrium? • Permits: Bridget might sell Abdul a fishing permit allowing him to catch not more than a given amount of fish, setting the highest possible fee for the permit consistent with Abdul being willing to fish under those terms. • Employment: Bridget might offer Abdul an employment contract under which Abdul would fish a given amount of time; the fish caught by Abdul would be Bridget’s. Abdul’s compensation would be a wage (paid in the fish caught by the two of them) which would be sufficient to offset the disutility of Abdul’s fishing time and the opportunity cost of his fishing (and therefore to satisfy Abdul’s participation constraint). R E M I N D E R While a single owner will take account of the costs of over-fishing and restrict fishing accordingly, the owner may also be a monopolist in selling the fish to others, and will restrict fishing even more than is socially optimal so as to sustain a high price of fish. We do not include the consumers of fish other than the owner and those fishing on the lake. But we will include them when we return to these effects of a monopolist in Chapter 9. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES Selling permits to fish To understand what Bridget will do as the owner of the lake, let us return to her utility function: Bridget’s utility uB (hA , hB ) = yB 1 B 2 (h ) 2 (5.52) In Equation 5.52, when Bridget was one of the two fisherman and could not charge anyone to access the lake, her production of fish equalled her consumption of fish (yB ). Now, though, as she will be charging Abdul to access the lake, we separate her production of fish, x(hA , hB ), from her consumption of fish, yB . When she is the owner, her consumption of fish equals her own production plus the fee she charges, F . Her utility therefore becomes: Bridget’s disutility Bridget’s production Owner’s utility z }| { z }| { 1 B 2 B A B A B u (h , h , F ) = x(h , h ) + F (h ) | {z } 2 (5.53) Bridget’s consumption Equation 5.53 tells us that Bridget, as the owner, now has three variables to determine not just one (her own fishing time) when she interacted with Abdul in the symmetric game. • her own fishing time, hB • Abdul’s fishing time, hA • the cost F to Abdul of the permit allowing him to fish hA hours. Here, the fee for the permit will be an amount of fish that Abdul would transfer to Bridget, but the idea easily extends to thinking about monetary payments rather than payments in kind like this one, since fish in this scenario are effectively money. Bridget, being self-regarding, will want to know: "what is the largest fee that I can charge Abdul?" Remember Abdul has the option of not fishing at all and and receiving a transfer of an amount yz of fish; this is his fallback position with associated utility uA z = yz . Agreeing to fish in Bridget’s lake means foregoing the fallback option, so uA z = yz is the opportunity cost of fishing. This is Abdul’s participation constraint, limiting how much Bridget can charge for the permit: the fee plus the opportunity cost of fishing cannot be larger than Abdul’s utility from fishing, uA (hA , hB ): Abdul’s utility from fishing Abdul’s participation constraint A A B u (h , h ) Permit fee + Foregone fallback F + yz Because Bridget would never consider charging Abdul less than she could, we can assume that equation 5.54 will be satisfied as an equality and so (5.54) 265 266 MICROECONOMICS - DRAFT (re-arranging the equation) we have: F Abdul’s participation constraint (PC) = uA (hA , hB ) yz (5.55) Bridget’s constrained optimization problem is now to vary hA and hB to maximize her utility from fishing plus the fee she charges Abdul, or: Maximize her total utility: uB (hA , hB , F ) subject to Abdul’s PC: F = x(hA , hB ) + F = uA (hA , hB ) 1 B 2 (h(5.56) ) 2 yz (5.57) We can use the expression for F that is equation 5.57 to replace the F in equation 5.56, so that now Bridget’s objective is to chose hA and hB to: Maximize: uB (hA , hB ) = x(hA , hB ) 1 B 2 (h ) + uA (hA , hB ) 2 y(5.58) z B Once we have found the hA i and hi that maximize 5.58, we can insert those values of hA and hB into equation 5.57 to determine F , the cost of the permit to charge Abdul. Comparing equation 5.58 with equation 5.33 we can see that Bridget maximizes the same quantity that the Impartial Spectator maximized, namely the sum of the utilities of the two, except here yz is subtracted. But because yz is a constant (112 pounds of fish in our numerical examples) the solution of these two optimizing problems – the values of hA and hB chosen – must be the same. B This means when Bridget is the owner the hours worked, hA i and hi will be equal, ten hours each in our numerical example – but the levels of utility realized will be maximally unequal. Abdul will get exactly 112, his fallback position, and Bridget will get 188 as shown by point b in Figure 5.14. Because the allocation was determined by Bridget maximizing her utility subject to a constraint on Abdul’s level of utility (that is the participation constraint) it has to be Pareto efficient, by definition. If Bridget as the owner implemented a plan in which she reduced Abdul’s work hours and increased her own, she would obtain an allocation on the utility possibilities frontier at point b0 . HowB ever, if she implements the optimal number of hours (hA i = hi = 10) at point i and has Abdul pay her for the right to fish with a fee, then the total rents available are 300. She will charge a fee such that Abdul will receive just a bit more than his fallback, uA z = 112 and she will get 300 112 = 188. This corresponds to a movement along the blue line with slope = 1 from point i to point b. The blue line therefore indicates movements from point i to alternative feasible trades when the fishermen are able to trade fish between them as payments. The economic reason why the result is Pareto-efficient follows directly from the fact that Bridget knew in advance that she would capture all of the feasible C O O R D I N AT I O N F A I L U R E S slope = − 1 A's PC 225 b B's utility, uB 188 178 bʹ i 150 B's PC 112 z 112 150 A's utility, uA 225 rents. Given that fact she had every interest in making the total rents as large as possible. Had she chosen a Pareto-inefficient outcome she would have foregone the opportunity to make herself better of without making Abdul worse off (than his participation constraint). But why wouldn’t Bridget select hA = 0, and have exclusive access to the lake? The reason is that the marginal cost of compensating Abdul’s fishing time is very small when he is not fishing much, or at all. So it is to Bridget’s advantage to let Abdul fish in the lake and pay her for the privilege, rather than doing all the fishing herself. Employing others to fish Instead of issuing a permit, Bridget might hire Abdul to work for her. Employment differs from the permit system in that when Bridget employs Abdul, she owns all of the fish caught by Abdul, but must devote some of this to paying a wage w to Abdul sufficient to satisfy his participation constraint. From our reasoning in the permit case we know that the participation constraint will be satisfied as an equality. This allows us to use the fact that the total wage paid (w) must offset Abdul’s disutility of fishing time and the opportunity cost of fishing (namely his fallback option, yz that he gives up if he fishes) or: Abdul’s PC as an employee w= ( hA ) 2 + yz 2 (5.59) & INSTITUTIONAL RESPONSES 267 Figure 5.14: Payments in fish takes the fishermen to allocations outside of the feasible set. The blue line with slope -1 shows the allocations of utility that are possible if the two fish at the socially optimal times indicted by point i follow a transfer of fish from on to the other. The slope is -1 because the opportunity cost of, say Bridget having a kg more fish, is that Abdul has one kg less. If Bridget as the owner implemented a plan in which she reduced Abdul’s work hours and increased her own, she would obtain an allocation on the utility possibilities frontier at point b0 . However, if she implements the optimal number of hours (hAi = hBi = 10) at point i and has Abdul pay her for the right to fish with a fee, then the total rents available are 300. She will charge a fee such that Abdul will receive just a bit more than his fallback, uAz = 112 and she will get 300 112 = 188. This corresponds to a movement along blue trade line with slope = 1 from point i to point b. 268 MICROECONOMICS - DRAFT Bridget then must choose hA and hB to maximize her utility: uB (hA , hB , w) = xA (hA , hB ) + xB (hA , hB ) = A’s catch + B’s catch 1 B 2 (h ) 2 B’s disutility w A’s wage(5.60) Then using equation5.59 for Abdul’s wage, what Bridget maximizes when she employs Abdul is: uB (hA , hB ) = hB (a b (hA + hB )) (hB )2 + hA (a b (hA + hB )) 2 ( hA ) 2 2 yz (5.61) Equation 5.61 can be understood as follows: • it is identical to what Bridget maximized in the permit case, namely equation 5.58, and • identical to what the Impartial Spectator maximized, namely, equation 5.33 minus the constant yz . In both the permit and the employment cases, the outcome is Pareto efficient, but Abdul gains only an amount equal to his disutility of fishing time plus the opportunity cost of his fishing at all. The allocation proposed by the Impartial Spectator and that implemented by Bridget as owner of the lake does not differ in the fishing times of each, or the degree of exploitation of the fishing stock. In this sense private ownership of the lake has addressed the Pareto-inefficiency of the over-exploitation of the lake as a common property resource. The only difference is that in the private ownership case there is transfer of rents (amounting to 88 pounds of fish in both cases) from the non owner to the owner: • In the permit case the transfer took the form of the fee for the permit to fish (88) that Abdul paid to Bridget. • In the employment case the transfer occurred because Bridget owned all of the fish that Abdul caught (200), 88 pounds of which she retained for her own consumption after paying him the wage (112). This is general feature of social coordination problems. When one actor is sufficiently powerful to maximize their utility subject to the participation constraint of other actors, a Pareto-efficient allocation will result, and the powerful actor will get all of the economic rents. We have already seen this pattern in the TIOLI power scenario in Chapter 4. Checkpoint 5.11: Wages vs. Permits Refer to M-Note ??. Assume the same values for the parameters of a and b . C O O R D I N AT I O N F A I L U R E S a. What wage would Bridget pay Abdul if Abdul’s fallback was zero? b. What wage would Bridget pay Abdul if Abdul’s fallback was to have a job working as an administrator for a utility of uA z = 100 rather than working for Bridget? 5.13 Community: Repeated interactions and altruism Here and in previous chapters we have used two-person games to represent economic interactions among a very large number of people. But some of our interactions are with small numbers of people, for example, in our neighborhoods, families and workplaces, and even, in some cases, in exploiting a local common property resource like a forest or fishery. These small communities often address coordination problems in ways not possible when the number of people interacting is very large. This is possible because members of small communities: • often have information about each other that is not available to governments or private owners who are not part of the community; • interact repeatedly with each other repeatedly so that there are opportunities to retaliate against members who violate social norms or informal agreements; and • often care about each other, and these social preferences can reduce conflicts of interest (as we saw in the previous chapter) and can provide the basis for addressing coordination problems. These characteristics of small communities give them capabilities in solving coordination problems that are unavailable to purely government- or marketbased approaches. As we have seen in Chapter 2, public goods experiments show that people are willing to punish fellow group members whose behaviors violate norms, even when inflicting the punishment is costly to the punisher. Lets see how a small community of fishing people – illustrated by Bridget and Abdul – might address the over-exploitation of the common property resource. Repeated interactions In the one-shot games we have introduced so far the strategies available to the players are limited: select some amount of fishing hours. One way to make the game more realistic is to let the interaction be repeated over possibly many periods with the same players. Then more complicated strategies are possible, even if in every period there are just two actions one can take, for example, fish ten hours or fish twelve hours. Importantly, strategies can now be conditional on what the other player has done in previous play. & INSTITUTIONAL RESPONSES 269 270 MICROECONOMICS - DRAFT Bina 140 d 140 144 ● 156 ● b 144 c (a) Stage game One strategy might be play the strategy that would implement the social optimum (fish 10 hours) in the first round and on the next and successive rounds of the game, play whatever the other player played on the previous round. This strategy is called "nice tit for tat": nice because it begins with a strategy that could be mutually optimal, but "tit for tat" because it punishes the other player if she takes the over-fishing option. Consider a repeated game between Abdul and Bridget with the following properties: • Actions: In each period of the game, they may fish either 10 hours, the socially optimal amount or 12 hours, the over fishing level at the Nash equilibrium of the symmetric game in which they do not coordinate in any way. • Duration of the game: After every period that the game is played it is continued with some probability 0 < P < 1 We show in the Mathematics Appendix that this means that the expected duration of the game in number of periods is 1 1 P . • Payoffs: In each period the payoffs are given in panel a of Figure 5.15. The cell entries are from the analysis of their interaction we have carried out so far, with the parameters used in our numerical examples. Payoffs for the game are the sum of payoffs for each period the game is played. • Strategies. Each may choose either to fish 12 hours in every period of the game (called "Defect") or fish 10 hours in the first period of the game and every subsequent period until the other fishes 12 hours, in which case fish Grim Trigger 156 Defect 150 150 a Grim Trigger 12 Hours Aram 10 Hours 12 Hours Aram 10 Hours Bina Defect 1500 1452 ● 1500 1436 1436 1440 ● 1452 1440 (b) Repeated Interaction Figure 5.15: Repeated interactions can convert a one-shot Prisoners’ Dilemma into an Assurance Game allowing or coordination on a socially optimal allocation. Panel a is the payoff matrix for the one-shot (stage) game between Bridget and Abdul, that is played once only. Inspection of the payoffs shows that it is a Prisoners’ Dilemma. You can confirm using the circle and dot method introduced in Chapter 1 that each player fishing 12 hours is a Nash equilibrium (the circles and dots show that this is also a dominant strategy equilibrium. Panel b gives expected payoffs for the game, if at the end of each period with probability P = 0.9 the game is played again (with the same payoffs per period as shown in panel a). In this case the circles and dots indicate that the repeated game has two Nash equilibria: the Nash equilibrium of the one-shot game with payoffs to each of 1440, and the socially optimal allocation with payoffs 1500. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 271 12 hours as long as the game lasts. This strategy is called "Grim trigger": the term trigger is used because on act of defection by the other sets of a a punishing defection by the actor, it is grim because the defections go on as long as the game lasts. How will this repeated game be played? We will assume that both players are entirely self regarding. You can see that the single period payoff matrix – called the stage game – has the structure of a prisoners’ dilemma. The repeated game will continue following each period with probability P, which for concreteness we set as P = 0.9. This means that the expected duration of the game is 1 10.9 = 10 periods. If the game were played among total strangers, it is unlikely that it would be repeated with such a high probability. But if the two players are neighbors or co-workers they are very likely to continue interacting. The payoffs in the repeated game are derived as follows. Playing Defect against Defect The payoff will be the payoff (144) of the mutual defect option of the stage game (fishing 12 hours) for as long as the game continues (10 periods) so the payoff is 1440. • Playing Grim Trigger against Grim Trigger : The payoff will be 150 (the stage game payoff to fishing 10 hours) for as long as the game lasts, or 1500. • Payoff to playing Grim Trigger against Defect: In the first period, the Grim Trigger player fishes 10 hours, while the defector fishes 12 hours, and so receives 140 that period; and then he defects in the next period (should it occur) and until the game ends. The probability that the second period happens is P and if it does it can be expected to continue for 10 more periods so the payoffs from period two to the end of the game are 0.9 ⇥ 10 ⇥ 144 or 1296 which, adding the first period’s payoffs totals 1436. • Payoff to playing Defect against Grim Trigger : This is calculated exactly is in the case immediately above. The defector gets 156 in the first period and then the mutual defect payoff as long as the game lasts, totalling 1452. Looking at panel b of Figure 5.15 you can see that mutual Defect is still a Nash equilibrium in the repeated game: it is a best response to both Defect and Grim Trigger, as the circles and dots show. But in this case the repeated game is not a prisoners’ dilemma: the best response to Grim Trigger is not Defect, but Grim Trigger itself. And so if the two were to coordinate on fishing at the socially optimal level (10 hours) and had decided to play Grim Trigger they would continue doing so until the game ended. This means the what was Prisoners’ Dilemmas if played as a one shot can M - C H E C K We show in the appendix that the expected duration of an interaction is the inverse of the probability that at the end of a period the interaction will be terminated. 272 MICROECONOMICS - DRAFT become an Assurance Game if played repeatedly if the game is repeated with a sufficiently high probability. What this means is that entirely self-regarding actors acting independently and without government regulation can escape the prisoners’ dilemma. You can confirm using the circle and dot method introduced in Chapter 1 that each player fishing 12 hours is a Nash equilibrium (the circles and dots show that this is also a dominant strategy equilibrium). Panel b gives expected Bina payoffs for the game, if at the end of each period with probability P = 0.9 the game is played again (with the same payoffs per period as shown in panel a). 10 Hours 12 Hours 1440, and the socially optimal allocation with payoffs 1500. M-Note 5.13: Cooperation without agreements in a repeated game The key to how game repetition converts a Prisoners’ Dilemma stage game into an Assurance Game that can implement the socially optimal level of fishing is that Defect should not be a best response to Grim Trigger. This requires that the payoff to playing Defect against Grim Trigger should be less than the payoff to playing Grim Trigger against itself, or (using he letters in the payoff matrix of panel a): b+c ✓ P 1 P ◆ < a 1 P which, re-arranged to isolate the P gives us: P> b b a c This means that(in the one shot game) the probability that the game will be continued after each round must be greater than the payoff advantage of defecting on a cooperator (b a) relative to the payoff advantage to coordinating on 10 hours rather than 12 hours (b c). This means that repeating the game is more likely to result in the Pareto superior symmetric outcome (both fishing less) if • the incentive to exploit the cooperation of the other is less • the joint benefit of mutual restricting fishing hours is more and if • the interaction will be repeated with high probability For the payoffs in 5.15 a this condition is satisfied for any P > 0.5 Social preferences: Altruism The fact that the community of fishermen is small means two important facts about their context are likely to hold. First, people in small communities can more easily access information about one another and engage in repeated interactions as the basis for retaliation against those community members who defect. Second, small communities are also often the basis of the people having a concern about each others’ well-being. such as altruism, fairness concerns, or reciprocity.8 To see how social preferences might help solve coordination failures, suppose that in choosing an action each participant puts some weight on the utility of Aram equilibria: the Nash equilibrium of the one-shot game with payoffs to each of 12 Hours 10 Hours In this case the circles and dots indicate that the repeated game has two Nash a a b d c d b c Figure 5.16: Payoffs for the one-shot Prisoners’ Dilemma Game shown in Panel a of Figure 5.15 repeated here to accompany M-Note 5.13. For both players: b > a > c > d . C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 273 the other as in Equation 5.62, so that Abdul and Bridget have social preferences like those we introduced in Chapters 3 and 4. Because we have used uA and uB to refer to the utility each gets from fishing, we now introduce vA and vB , which include their concern for the other’s well-being. Because they are other-regarding, their evaluation of the outcomes they believe their actions will produce are based on these social preference utility functions, vA and vB . Altruistic A: vA (hA , hB ) = Own utility + l Other’s utility = uA + l uB Altruistic B: vB (hA , hB ) = uB + l uA We show in the M-Note that the best-responses maximizing these utility functions are: hB ( hA ) = B’s best response : (a ( 1 + l ) b hA ) 1 + 2b M-Note 5.14: The best-response function of altruistic people Now, Abdul cares about Bridget. His utility function is: vA (hA , hB ) = uA + l uB = hA (a b (hA + hB ) ✓ 1 A 2 (h ) + l hB (a 2 b (hA + hB )) 1 B 2 (h ) 2 ◆ To obtain his best response function, we differentiate Abdul’s utility functions with respect to fishing time and set it equal to zero: ∂ vA ∂ hA (1 + 2b )hA hA = a 2b hA = a (1 + l )b hB a ( 1 + l ) b hB 1 + 2b = b hB hA l b hB = 0 Each takes account of a fraction, l , of the external effect that they have on the other person. Concern for the well-being of other people thus might at least partially substitute for a tax or government regulation in alleviating the social coordination failure: when people care for each other they are willing to internalize the external effect of the cost they impose on other people. What level of concern for the other would implement the Pareto-efficient outcome? In order for altruism to implement the social welfare maximizing allocation proposed by the Impartial Spectator, these altruism based best response functions would have to to mimic those of the Impartial Spectator in Equations 5.40 and 5.41. This means that the (1 + l ) would have to take the value of M - C H E C K Notice here that 0 l 1 rather than 0 l 12 in Chapters 3 and 4. This is because in those chapters the utility functions are Cobb-Douglas whereas in this chapter we are adding up the utility functions to display altruism (making these functions Cobb-Douglas would be very tough mathematically!). 274 MICROECONOMICS - DRAFT 2, meaning that we would have to have l = 1. Each fisherman would have to be the perfect altruist that we defined in Chapter 2, caring as much for the other person as she does for her self (namely l = 1). The difficulty of sustaining this level of altruism may suggest why most successful communities do not rely entirely on good will, but supplement it with mutual monitoring and punishment for transgression of norms. Checkpoint 5.12 1. Refer to Equations 5.50 and 5.44. Why do we need l fully Pareto-efficient outcome? Explain. = 1 to implement a 2. What would it mean if l < 1? If people aren’t only altruistic towards each other in a community, what else might they do to affect the utility others receive if they break social norms by over-fishing? 3. What would it mean if l > 1. Would the outcome be Pareto-efficient? Why or why not? 5.14 Application: Is inequality a problem or a solution? Recall that there two standards that we use to evaluate policies to address coordination failures: • Does it result in a Pareto-improvement over the status quo, that is, does it improve efficiency by making at least one of the participants better off and none worse off? • Is the resulting allocation more fair than the status quo, that is, are the rents (improvements over the status quo) that the players receive fair? In some cases the two objectives can be jointly realized; in others they are in conflict. The distribution of the economic rents resulting from coordination depend on the particular transformation of the game which makes coordination possible. Conflicts may arise about how best to address the coordination problems that people face: some participants may prefer an inefficient solution to the allocation problem because they get a larger share of the economic surplus at a Pareto-inefficient outcome. Unequal solutions to local social coordination problems are generally based on the disproportionate wealth or power of one of the fishermen. It is easy to see that if one of the fishermen has a much larger net than the others and so can be assured of catching most of the fish, then his best response will approximate the allocation of a single owner of the lake. In this case, inequality in wealth among the fishermen would lessen the coordination failure. Important inequalities may exist even among otherwise identical fishermen. To see this consider two possibilities: E X A M P L E Even the most utopian, such as the contemporary Amish or Hutterite communities in the U.S., supplement altruistic action with social sanctions and monitoring of peers (which is easier to do in small communities). C O O R D I N AT I O N F A I L U R E S uB2 uBn uBf B's BRF n hBN = 12 h = 11.8 uAn ● B's hours, hB BF ● f uAf tB ● Pareto−efficient curve uA3 A hBtA = 9.4 t & INSTITUTIONAL RESPONSES 275 Figure 5.17: First-mover advantage: Fishing time-setting power. In the figure, Abdul is the leader with fishing time-setting power (first-mover power) and Bridget is the follower. Abdul doesn’t have enough power to make a TIOLI offer, but he can make a credible commitment of his own fishing time such that Bridget will have to adapt in choosing her fishing time to Abdul’s fishing time. Abdul, when choosing his fishing time, takes Bridget’s best-response function as his incentive compatibility constraint. He maximizes his utility subject to satisfying her incentive compatibility constraint, finding the point at which his indifference curve is tangent to her bestresponse function as occurs at point f where Abdul exerts fishing time hAS and Bridget exerts fishing time hBS and the two fishermen obtain utilities uA2 = uAS and uB1 = uBS . This Stackelberg or Leadership outcome is contrasted with the Nash equilibrium outcome of the simultaneous interaction where Bridget had higher utility uB2 = uBN and Abdul had lower utility uA1 = uAN . A's BRF ● hAtA = 10.55 A hAN = 12 hAF = 12.9 A's hours, h 1. Take-it-or-leave-it (TIOLI) power: When one player’s social position is such that they have substantially more power than the other making a TIOLI offer of both their own and the other’s fishing time, then they may implement an allocation where they obtain all the economic rents and the other remains on their participation constraint. As you already know from the previous chapter, the outcome is Pareto-efficient. 2. First-mover advantage: When one player’s social position is such that they can credibly commit to a fishing time such that the other must simply respond, they obtain more of the economic rent than the other player, but not as much as if they had TIOLI power. This is similar to the price-setting power in Chapter 4. The outcome is Pareto-inefficient. We start with first-mover advantage: the power to commit to one’s own fishing time. First-mover advantage: fishing time-setting power Suppose that Abdul can announce a level of fishing time and commit to it in such a way that Bridget understands that nothing she can do will alter Abdul’s fishing activity. Bridget will then select her level of fishing to maximize her utility given what Abdul has committed to. In this situation, Abdul is the firstmover and has fishing time-setting power similar to the price-setting power (as in Chapter 4). Economists call Abdul in this situation the Stackelberg leader. H I S TO RY Heinrich von Stackelberg (19051946) used this model to represent pricesetting among duopolists (two firms in a market), which is why this type of model is named after him. 276 MICROECONOMICS - DRAFT The big difference between Abdul having fishing time-setting power and our previous analysis is that the game is now sequential and the order of play matters: who gets to go first is important. How would Abdul decide what level of fishing time to commit to as the fishing time-setter? As the first-mover, he will begin by determining what the secondmover will do in response to each of the his actions, and then select the action that maximizes his own utility given the best-response function. The secondmover’s best-response function is the incentive compatibility constraint. Abdul maximizing his utility subject to Bridget’s incentive compatibility constraint is a simple but important change in the assumed behavior of the fishermen: Abdul now recognizes and takes advantage of the fact that by choosing various levels of fishing time he can affect the level of fishing time Bridget chooses. Abdul’s behavior is strategic because it takes account of Bridget’s R E M I N D E R In Chapter 4, when she had Price-setting power, A maximized her utility subject to B’s price-offer curve. B’s priceoffer curve was A’s incentive compatibility constraint, which is exactly what a bestresponse function is: a best-response function shows what action would be your best response to the action (e.g. price or fishing time level) the first mover commits to. reaction to his action. In this first-mover case, Abdul is constrained not by a given level of Bridget’s utility, but by Bridget’s maximizing behavior as given by her best-response function. As a result, the first-mover outcome will not be Pareto-efficient because Bridget’s indifference curve intersects Abdul’s indifference curves at point f in Figure 5.17. Because the indifference curves intersect, the marginal rates of substitution are not equal and therefore the allocation is Paretoinefficient. Similar to the analysis of price-setting power in Chapter 4, the fishing time-setting outcome is not Pareto-efficient, because when Abdul maximizes utility subject to Bridget’s best-response function, he does not fully internalize the external effect. Abdul’s first-mover advantage allows him to improve his position by comparison to the Nash equilibrium, in this case at the expense of Bridget whose outcome as second-mover is worse than the Nash equilibrium. Take-it-or-leave-it power Let us switch roles now and consider what would happen if Bridget had more power than Abdul. She has enough power to make Abdul a take-it-or-leave-it offer, specifying not only how much she would fish, but how much Abdul is to fish, too, along with the threat that should Abdul not accept the offer, then Bridget would simply fish at the level of the Nash equilibrium of the simultaneous move game. Because Abdul will refuse her offer if it is worse for him than the Nash outcome, Bridget must make an offer to Abdul that satisfies Abdul’s participation constraint. If she does so, the outcome will be Pareto-efficient. Referring to Figure 5.12, we can see the outcome that Bridget would implement if she had take-it-or-leave-it power over Abdul. In the case where Abdul’s fallback position is the Nash equilibrium, then his participation constraint is R E M I N D E R When either trader had TIOLI power in Chapter 4, the exercise of power led to a Pareto-efficient outcome. C O O R D I N AT I O N F A I L U R E S given by uA n . As a result, Bridget would find the allocation on Abdul’s participation constraint at which she would maximize her utility. In Figure 5.12, Bridget’s TIOLI offer is shown at point tB . At tB , her indifference curve uB 2 is tangent to Abdul’s indifference curve uB n , so the allocation she chooses for her TIOLI offer is Pareto-efficient, unlike when Abdul was the leader and could set a fishing-time for Bridget. If Abdul had TIOLI power over Bridget, he would implement the allocation at point tA and this allocation would also be Pareto-efficient. Summing up, our model shows that the effect on unequal power will always benefit the more powerful and may, but need not, result in a Pareto-inefficient outcome. Abdul’s fishing first-mover ability (his time-setting power), resulted in gains for Abdul, losses for Bridget, and increased over-exploitation of the lake. This resulted in greater inequality than the Nash equilibrium, but the outcome was not Pareto-comparable to the Nash equilibrium (one does better and the other worse at the Nash). Positive effects on Pareto efficiency occurred when Abdul had take-it-or-leaveit power and the outcome was Pareto efficient, but probably regarded as unfair by Bridget or an Impersonal Spectator. The model is hypothetical but the problem is not. Checkpoint 5.13: Numerical TIOLI power Using the values of a = 30 and b = 12 : 1. Find the take-it-or-leave-it offer that Abdul would make to Bridget at point tA . How many hours would Abdul fish? How many hours would Bridget fish? 2. What would Abdul’s utility be at the TIOLI offer? What woudl Bridget’s utility be at the TIOLI offer? Evidence from field studies A field experiment among forest commons users in rural Colombia underlines how inequality may be an impediment to achieving more satisfactory outcomes through coordination. Juan Camilo Cardenas implemented common pool resource behavioral experiments among villagers who rely for their living on the exploitation of a nearby forest.9 So the subjects in the experiment were in real life playing the same game that the experimenter invited them to play. In Cardenas’s game, the subjects choose to withdraw a number of tokens from a common pool (these represented exploitation of the common property resource), and after all subjects had taken their turn the tokens remaining were multiplied by the experimenter and then distributed to the players, the tokens then being exchanged for money. (This is similar to the Public Goods Game experiment in Chapter 2 except that subjects decide how much to & INSTITUTIONAL RESPONSES 277 278 MICROECONOMICS - DRAFT withdraw rather than how much to contribute to the pool). For an initial set of rounds of the game, no communication was allowed. But in the final rounds of the game, subjects were invited to converse for a few minutes before making their decisions. Cardenas expected that communication would reduce the level of withdrawals from the common pool (as has been the case in similar experiments) despite the fact that it does not alter the material incentives of the game. Communication was indeed effective among groups of subjects with relatively similar wealth levels (measured by land, livestock and equipment ownership); their levels of cooperation increased dramatically in the communication rounds of the experiment. But this was not true of the groups in which there were substantial differences in wealth among the subjects. In one group, one of the wealthiest subjects tried in vain to persuade his fellow participants (who in real life were his tenants and employees) to restrict their withdrawals to the socially efficient amount, in order to maximize their total payoff. But the wealthy subject’s advice fell on deaf ears. “I did not believe Don Pedro,” one of the less well-off women in his group later explained, “I never look him in the face.” She was right not to trust him: Don Pedro (not his real name) had withdrawn the maximal amount despite his contrary advice to the other players. This is not an isolated example. • A study of water management in 48 villages in the Indian state of Tamil Nadu found lower levels of cooperation in villages with high levels of inequality in landholding. Moreover, lower levels of compliance were observed where the rules governing water supply were perceived to be chosen by the village elite. • A similar study of 54 farmer-maintained irrigation systems in the Mexican F AC T C H E C K In a recent study of participation in church, local service and political groups, as well as other community organizations providing local public goods by Alberto Alesina and Eliana La Ferrara found that participation in these groups was substantially higher where income is more equally distributed, even when a host of other possible influences are statistically "held constant."10 state of Guanajuato found that inequality in land holding was associated with lower levels of cooperative fishing time in the maintenance of the field canals.11 In other cases, inequalities based on traditional hierarchical have made a positive contribution. • Another study of Mexican water management, for example, found that increased mobility of rural residents undermined the relationships that had been the foundation of a highly unequal but environmentally sustainable system of resource management.12 • And in the port of Kayar, on the Petite Côte of Senegal, a cooperative fishing time to limit the catch (to support higher prices, not to protect fishing stocks) owed its success in part to the leadership of the wealthy local F AC T C H E C K Social differences among commons users affects outcomes in other ways. The fishing agreement in Kayar was threatened by conflicts between locals and outsiders using differing technologies, and other attempts to limit fishing failed due to the indebtedness of fishermen to fish sellers (who opposed the limits) and because the wives of many of the fishermen were fish sellers. C O O R D I N AT I O N F A I L U R E S traditional elite of elders.13 5.15 Over-exploitation of a non-excludable resource We stated at the beginning of the chapter that we would illustrate the problem of the common pool resource problem by the example of just two people. This was despite the fact that, as a non-excludable resource, there would be no limit on the number of people who could, if they wished, fish on the lake and compete with Abdul and Bridget for the available stock. We have so far studied just one of the two aspects of the coordination failure resulting in over-fishing: the fact both Abdul and Bridget fished more hours than was Pareto-efficient. They both could have been better off had they been able to agree to fish less. Now we introduce the second aspect of the problem: many more people could fish the lake. Because the lake is a common pool resource, its nonexcludablity property of a means that there is open access. How many people would use the lake? To answer the question we add more context to the initial problem. Abdul and Bridget are part of a large community of people who may make their livelihood fishing on the lake, or if not that, then doing some other kind of work yielding a utility of uz (their fallback option). People will decide to make their living fishing as long as the utility they gain exceeds uz . In Figure 5.18 using the same values for the parameters as in the other numerical examples in this chapter, we calculate the maximum utility that could be attained by one person fishing alone, two (as in the case of Bridget and Abdul), three, and so on up to 11. However many there are fishing, they will receive the same utility in the Nash equilibrium because they are identical, and we have so far assumed that none has any advantage in bargaining with the others. The height of each bar is the utility attained. When there are just two people fishing as in our previous examples involving Abdul and Bridget each receives a utility of 144, as we found in M-Note 5.6. The more people that fish in the lake the lower the utilities each of them receive will be. When there are ten they all have a utility of 21.3, barely greater than their fallback options. Now think about some other member of the community who is not currently fishing but is thinking of doing so. Those fishing are doing better than the fallback options. But if the 11th person decided to fish they would all receive a utility of 18.4 (including the new fisherman). That is, they would all receive less than their fallback options. So the eleventh person would decide not to fish. & INSTITUTIONAL RESPONSES 279 280 MICROECONOMICS - DRAFT 240 220 200 180 Utility 160 140 120 100 80 fallback option = 20 60 40 Figure 5.18: The dynamics of over exploitation of a common property (non-excludable) resource. All of the people who might fish on the lake have the same utility functions as Abdul and Bridget with the values of a = 30 and b = 12 . The height of the bar for a given number on the x axis is the utility of each of the fishermen when there are the indicated number fishing on the lake. The fallback utility is uz = 20. You can see from the figure that if the lake is a common property resource, so that no fisher can be excluded, the Nash equilibrium number fishing on the lake is 10 with each receiving a utility of 21.3. If the 11th person fished on the lake, she – and all of the rest of the fishermen there – would receive a utility of 18.37, that is, less than their fallback option). The mathematics on which this figure is based are shown in M-Note 5.15. 20 0 1 2 3 4 5 6 7 8 Number of people, n 9 10 11 Generalizing from this example, the Nash equilibrium number of people fishing is the largest whole number of people fishing such that the utility that those fishing receive is greater than or equal to the utility they would receive at their fallback option. As a result, the Nash equilibrium of this game is that we have: • nN = 10 the number of people fishing and • hN = 4.62 the number of hours each of them works. We use the N superscript for each of these quantities because both are Nash equilibria (but under different rules of the game): • nN = 10 is a mutual best response because none of those fishing could do better by not fishing, and none of those not fishing could do better by fishing, and • given that ten people are fishing, then hN = 4.62 is also a mutual best response because for each person fishing this is a utility-maximizing choice of hours, given the hours that everyone else is fishing. The Nash equilibrium is Pareto-inefficient for two reasons: too many people are fishing too many hours each. Just as was the case with Abdul and Bridget, if each fished a little less they all would be better off. And if fewer of them fished, all ten of them could be better off. Figure 5.18 shows that if 3 people fished they would each have a utility of 100. We call people who are already doing an activity, such as fishing or owning a firm, incumbents. We therefore call the existing fishermen, the incumbents or incumbent fishermen. Suppose this was the case, and that the incumbents could somehow agree to bribe the other 7 not to fish. Notice we have just changed M AT H - C H E C K The equilibrium number of people fishing must be a whole number because the entry of a "fractional fisherman" would not make much sense unless we allowed people to split their day between fishing and the fallback option, which we do not. C O O R D I N AT I O N F A I L U R E S the rules of the game to allow the incumbent three to coordinate. The incumbent three would have to give the other 7 potential fishermen an amount of fish sufficient that each would be as well off as their fallback. This amount would be uN uz or 21.3 20, or 1.3 each. The total payments by the three to the other seven would be 7 ⇥ 1.3 = 9.1, leaving each of the three better off (each receiving 300 9.1)/3 = 97). If the incumbent three increased the ‘bribe’ just a little bit then all 10 would be better of than at the Nash equilibrium with open access. M-Note 5.15: Nash equilibrium number of people fishing Because access to the lake is open to all, the number fishing there will be the largest whole number (because we cannot have fractions of people fishing) such that the utility of those fishing is equal to than the fallback option (their utility if they are not fishing in the lake), which is uz = 20. To determine this number, we first derive hN (n) the hours of fishing that each will do as a function of the numbers fishing, and use this result to determine uN (hN (n)) the utility of those fishing as a function of how many there are. To determine hn (n) we study the utility maximization problem of person 1: max u1 h1 = = n 1 2 h 2 1  hi ) h1 (a b h1 (a b h1 i=1 b n  hi ) i=2 1 2 h 2 1 (5.62) To find the hours of fishing that maximizes the utility of person 1 we differentiate equation 5.62 with respect to h1 , and set the result equal to zero. This gives us the first order condition: a marginal benefit = a n b 2b h1  hi h1 = 0 b  hi = h1 = marginal cost i=2 2b h1 n i=2 (5.63) Rearranging Equation 5.63 we get person 1’s first order condition giving the utility maximizing amount of fishing time: (1 + 2b )h1 h1 = a b n  hi i=2 = a b Âni=2 hi 1 + 2b (5.64) All face the same first order condition so in the Nash equilibrium all fish the same amount of hours: h1 = h2 = . . . = hN . Equation 5.64 becomes: a b (n 1)hN 1 + 2b = a b (n b hN = a (1 + b + b n)hN = a N = hN = (1 + 2b )hN hN + 2b hN + b nhN h 1)hN a 1+b +bn As in the rest of the chapter, we let a = 30 and b = 12 . You can verify that, if n = 10, then & INSTITUTIONAL RESPONSES 281 282 MICROECONOMICS - DRAFT 1 + b + b n = 6.5 and so hN (10) = uN (10) = = 30 6.5 = 4.61. The utility of each fisher would be: 4.61 ⇥ (30 21.30 1 ⇥ 10 ⇥ 4.61) 2 1 ⇥ 4.612 2 (5.65) If one more enters takes up fishing so that n = 11, then the hours of fishing would be hN (11) = 30 7 = 4.29.The new utility of each fisher would be now: uN (11) = = 4.29 ⇥ (30 18.37 1 ⇥ 11 ⇥ 4.29) 2 1 ⇥ 4.292 2 (5.66) But this ( n = 11) cannot be a Nash equilibrium, because everyone – including the new entrant – would then be worse off than with the fallback option, uz = 20. So the 11th person would not enter (or if she did, others would leave). So the Nash equilibrium is nN = 10. This is illustrated in Figure 5.18. 5.16 The rules of the game matter: Alternatives to over-exploitation The new rules of the game allowing the incumbent three to bribe the others is just a thought experiment demonstrating that the Nash equilibrium with open access is not Pareto efficient. But commonly observed real-life rules of the game – like inequalities in bargaining power, cooperative management of the lake, or private ownership – could also address the over-fishing problem. Checkpoint 5.14: Pareto efficiency and open access a. Explain why open-access Nash equilibrium outcome with 10 fishermen is not Pareto efficient. What alternative, if any, is Pareto superior to it? b. Given your reasoning for a., do you think there are alternative outcomes that are Pareto superior to, say, three fishermen bribing the other 10 not to fish? Explain what the dynamics for the situations you describe would be? How many fishermen? How many hours spent fishing? And so on. TIOLI bargaining power To see that the institutions governing the interactions among them matter, think about the case in which one of the ten people fishing on the lake has the power to make a take-it-or-leave-it offer to the rest. Here we have changed the rules of the game by giving one of the fishermen TIOLI power (which allows a kind of coordination). But the lake is still open access, so there are ten fishermen there. The one with bargaining power – suppose it is Abdul – can now say to the others "each of you will fish x number of hours, and I will fish as many hours as I wish." This is the “take it" part of the offer. The "leave it"" part is: "and if you refuse, then I will return to fishing 4.61 hours." That is, return to the former Nash equilibrium hours that occurred when there was no coordination among the fishermen. C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES Open access, n=10 total hours: 46.15 283 A's Utility Others (not A) h = 4.62 h = 1.14 A has TIOLI power,n=10, total hours: 21.56 h = 11.3 Cooperative, n=10, total hours: 27.27 h = 2.73 h=6 A owns the lake, n=4, total hours: 24 h=6 Cooperative, n=4, total hours: 24 h=6 0 50 100 150 Utility The other fishermen would know that without coordination the best they could do is to all fish 4.61 hours, gaining a utility of 21.3. This is the others’ fallback option to the TIOLI offer. If accepting Abdul’s offer made them worse off than their fallback they would refuse, and just fish 4.61 hours. This is their participation constraint; if it is violated – so that they would receive a utility of less than 21.3 – the others will not accept ("participate in") Abdul’s offer. Abdul would know, therefore, that he needs to find the hours of all 10 of them (his and the rest) that maximizes his utility subject to the participation constraint on the minimum utility the others can receive. Figure 5.19 shows the offer Abdul would make, and the utility that he and the others would experience. The first row of the figure shows, from the previous figure, the result for the unlimited access case without coordination. When Abdul has TIOLI power the other fishermen work fewer hours (just 1.14 each, rather than 4.62 before), but get exactly what they had under the uncoordinated open access case. This is so because that level of utility – 21.3 – is the participation constraint on what Abdul can offer them. Abdul himself works 11.30 hours and enjoys utility equal to 153.2. Notice that the total number of hours is reduced sharply compared to the uncoordinated Nash equilibrium: from 46.15 to 21.56 hours. This reduction in total hours is the reason why the others are able to fish less but still attain the same utility: they catch more fish in an hour due to the lesser total hours of fishing. 200 250 300 Figure 5.19: The rules of the game: Noncooperation, bargaining power, and ownership. The bars show the utility of the fishermen (it is identical for all fishermen in the first and third row). The numbers at the end of the bars show the hours fished, where hA is the fishing hours of the owner or person who has TIOLI power, h A is the fishing hours of the non-owner or people who do not have TIOLI power. 284 MICROECONOMICS - DRAFT A democratic fishing cooperative An entirely different set of rules of the game – a democratic cooperative of the fishermen – would implement a correspondingly different set of results. Suppose that none of the ten fishermen has any bargaining power advantage and that they jointly own the lake. They can decide jointly – democratically by unanimous consent – on the same number of hours that each of them will fish. To figure this out they would think in the same way the Impartial Spectator did when she maximized total utility. They will maximize the sum of their utilities because this will also maximize the utility of each fisherman. The result is shown on line 3 of Figure 5.19. When there are 10 fishermen, they would each fish 2.73 hours and attain utilities of 40.9 each. Because their utility as coop members is now double their fallback option, others who are experiencing the fallback utility of uz = 20 would wish to join the cooperative. But it might be difficult to persuade the members to admit others, as this would reduce the utility of the incumbent fishermen. Their total fishing hours (27.3) is substantially greater than under the TIOLI power of Abdul, and so is their total utility (409.1 compared to 344.9). We can conclude two things from this last fact: • Suppose Abdul still had TIOLI power. If the other fishermen could coordinate their actions, they could ’bribe’ Abdul to give up his bargaining power and join their cooperative; they could have offered him the 153.2 that he received under his TIOLI power and still be better off dividing up the rest of their utility (fish) amongst themselves. They would each receive (409.1 153.2)/9 = 28.4, far better than the 21.3 they had when Abdul had bargaining power. So shifting from Abdul holding TIOLI power to the democratic cooperative is a Pareto improvement. • The reason why this is the case is that under Abdul’s TIOLI power they were as a group under -exploiting the fishing stocks. Abdul forced them to do this because the less the other people fished the more fish he could catch, and that was the only way he could increase his utility. The reason why the cooperative’s decision results in a greater total utility than the TIOLI case is that the members of the cooperative were pre-committed to sharing the total utility equally. And so they each had an interest in making total utility be as large as possible. Things would have been very different if Abdul had had the power to take some of the fish caught by the others (as in the employment and fee cases we dealt with earlier). In this case he would have done the same as the cooperative. He would have implemented the fishing times that maximized total utility. And then he would have taken fish from the others, leaving them just C OOPERATIVE A cooperative is a business organization or other association whose members together own the assets of the organization; they share the income resulting from their activities and jointly determine how the organization will be run (possibly through the democratic election of a manager). C O O R D I N AT I O N F A I L U R E S enough fish so that they did not decide to stop fishing. The TIOLI allocation was inefficient because Abdul’s bargaining power was limited. The TIOLI case was not inefficient because Abdul had some bargaining power. It was inefficient because he did not have enough power. As you will suspect from the 2 person case studied earlier, the allocation would have been Pareto efficient if Abdul had had all of the powers of a private owner of the lake. Private ownership Under these new rules – private ownership of the lake – the lake is no longer a common property resource because ownership means that Abdul can exclude anyone he wishes from fishing. Abdul would make three decisions: • How many other fishermen should I allow to fish in the lake? That is, what is the total-utility-maximizing number of people who should fish the lake? • How many hours should I allow them to fish? • If I employ them, then what wage should I pay them? Or if I charge them a fee for fishing, how large should the fee be? You know how to answer the second and third questions from the case earlier in the chapter when Abdul was the owner with just one other person Bridget on the lake. The first question is similar to that asked in Figure 5.18 but the answer is very different. The number of people fishing the lake is no longer based on the fishermen’s own decisions about where they can make a better livelihood. This is not their decision to make. The owner determines the number of others so as to maximize his utility. How he would do this is shown on line 4 in Figure 5.20. Abdul will allow three other fishermen to access the lake (so that means n = 4 including himself). Going back to Figure 5.19 remember, there are now just four fishermen, not 10 as before. All four fishermen fish 6.0 hours with the owner receiving a utility of 300 and each of the others receiving the same utility as their fallback option, that is, 20. The last line in Figure 5.19 shows what happens if Abdul is not the owner and instead if all four of the fishermen were members of a democratic cooperative. The members of the cooperative would implement exactly the same allocation of work time as occurred under private ownership: 6 hours of work time each. But the distribution of utility would be radically different, each of the four would receive 90. & INSTITUTIONAL RESPONSES 285 MICROECONOMICS Owner's utility 286 - DRAFT Figure 5.20: Utility of the owner when the lake is privately owned. On the horizontal axis are the total number of people fishing in the lake, including the owner. So, for example, where n = 2 we have Abdul as the owner and there is one other person, Bridget, the case we analysed earlier in this chapter. The height of each bar is the utility gained by the owner of the lake when he can both determine how any people fish there and dictate the terms under which they will work (as long as they receive utilities superior to their fallback position). 320 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 0 1 2 3 4 5 6 7 8 Number of people, n 9 10 11 We know that each working 6 hours is the allocation that maximizes total utility. The reason why private ownership of the lake implements this outcome is that the owner is limited only by the participation constraints of the others, and this is a constant (their fallback position of 20). So he implements an allocation to maximize the total utility, from which he must subtract the amounts required to keep the three others "participating." In sum, we can say the following: • Open access leads to a Pareto inefficient over-fishing outcome in which all the fishermen receive the same utility. • TIOLI power implements to a highly unequal and Pareto inefficient underexploitation of the lake. • Private ownership implements a Pareto-efficient and highly unequal outcome. • A democratic cooperative implements a Pareto-efficient and equal outcome. 5.17 Conclusion In practice, none of the approaches to addressing the common property resource coordination problem could be expected to work perfectly as: • no government is likely to have the information about the people’s preferences, production functions, and fishing times necessary to implement Pareto-efficient fishing levels by fiat, or to design the optimal taxes that would achieve the same result. • private owners face some of the same problems due to lack of information C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES 287 and moreover, while private ownership of a small lake or other common property resource is conceivable, for may important common property resources is infeasible (owning the oceans or the atmosphere for example) or undesirable (think of the unaccountable power that a private owner of such a vast common property resource could wield.) • altruism towards close family and loved ones will lead us to take at least some account of the effects of our actions on their well-being, but we are less likely to know or care deeply care about how our actions affect total strangers or even yet unborn generations who will benefit or suffer the external effects of what we do. The conclusion is not that the approaches to addressing coordination problems introduced here are ineffective. Each approach can contribute to making economic outcomes more efficient and fair. The models we introduced simplify the rules of the game that regulate how we interact with each other in exploiting a common property resource, in contrast to the vast diversity and complexity of rules that we observe. What the models have done is not to represent the world as it is, but to identify key aspects of how the world works to provide a lens for understanding them better. Making connections Social interactions and external effects: The interactions that economists study include buying and selling in markets, but they also include nonmarket interactions, sometimes called ‘social interactions’ ("social" here means simply non-market). The social interactions studied here include an external effect: the fact that one person’s fishing reduces the catch of another, and this effect is not taken into account when each of the fishermen decide how many hours to fish. Public goods, common pool resources and club goods: All of these things have the property such that each person’s actions have external effects on others and in the absence of social preferences or policies that internalize the external effects, these are not taken into account when people decide how to act, resulting in outcomes in which some potential mutual gains remain unexploited. Policy: Government policies and institutions may be designed so that people take account of external effects when they act. An example, is a tax on fishing that imposes on each fisher the marginal costs that their fishing imposes on the other fisherman, inducing each to choose their hours of fishing as if they cared about these external effects as if it as the actor herself rather than the other who bore the costs. Property: Converting a common property resource into a privately owned F AC T C H E C K Elinor Ostrom and her colleagues’ field research in different parts of the world from Colombia to Switzerland uncovered twenty-seven different local rules for excluding others from access to common property resources. These were based on such things as residency, age, caste, clan, skill level, continued use of the resource, use of a particular technology, and so on.14 288 MICROECONOMICS - DRAFT resource may result in a Pareto-efficient outcome in which the owner captures all of the potential mutual gains (rents). Power: When a single person has all of the bargaining power and so can make a binding take-it-or-leave-it-offer, he or she implements an outcome in which there are no unrealized mutual gains, and all of these gains (rents) go to the powerful person. Lesser forms of power – to commit to a particular fishing time, to which the other must respond, for example – advantage the powerful and can result in inefficient outcomes. Mutual benefits from coordination and conflicts over their distribution Policies to address coordination failures differ in how the resulting rents are distributed; the resulting conflicts may make it difficult to agree on any policy. Inequality: Differences in wealth, political connections and other sources of power can be both a source and a consequence of inefficient and unfair outcomes among people facing coordination problems. In some cases, these differences can also mitigate the inefficiencies arising from coordination failures. Models and relevance: Models, we wrote in Chapter 3, are like maps – a simplified guide to the territory, not the territory itself. But the model of social interactions introduced here, though quite abstract can be directly applied to very concrete economic actions such as firms competing for customers and, suitably extended, can illuminate global social interactions and coordination problems such as climate change and the spread of epidemic diseases. Important Ideas utility disutility external effect common property resource problem private property coordination failure rivalness excludability interdependence Impartial Spectator altruism reciprocity symmetrical interactions asymmetrical interactions social identification TIOLI power time-setting power stackelberg leader fiat power government policy decentralized implementation tax employment wage permit participation constraint fallback incentive compatibility constraint best-response function binding participation constraint social preferences solution C O O R D I N AT I O N F A I L U R E S & INSTITUTIONAL RESPONSES Mathematical Notation Notation Definition h a b u() v() W w F t a, b, c, d l fishing times parameter regulating the productivity of fishing times external effect of fishing time on the other’s productivity utility function value function expressing an altruistic concern for the utility of another person Impartial Spectator’s social welfare function wage in the employment solution permit fee in the permit solution per unit tax in the government policy solution payoffs in the repeated interactions game extent of altruism (valuation of the other’s utility relative to one’s own) Note on super- and subscripts: A and B: people; N: Nash Equilibria; i: Paretoefficient outcome; F: outcome with a first mover. Discussion questions See supplementary materials. Problems See supplementary materials. Works cited See Reference List. 289 Part II Markets for Goods and Services 293 Enter the Royal [Stock] Exchange of London, that place more respectable than many a court; you will see there agents from all nations assembled for the utility of mankind. There the Jew, the Mohammedan, the Christian deal with one another as if they were of the same religion. There the Presbyterian confides in the Anabaptist, and the Churchman depends on the Quaker’s word. ... They give the name infidel only to those who go bankrupt. Voltaire, 1734, Lettres philosophiques, Melanges (Paris, 1961) pp 17-18 When you hear the word "market" what other word do you think of? "Competition" probably is what came to mind. And you would be right to associate the two words. But you might have also come up with "cooperation". That is what impressed Voltaire about the London stock market: mutually advantageous interactions, even among total strangers "from all nations assembled for the utility of mankind." Markets allow us, each pursuing our private objectives, to work together producing and distributing goods and services in a way that, while far from perfect, is in many cases better than the alternatives. Markets accomplish an extraordinary result: unintended cooperation on a global scale, although often with a highly unequal distribution of the benefits. To better understand what markets do and how they work, begin with two workaday facts: We acquire skills as we produce things and,for this and other reasons, producing a lot of the same thing is often more effective in terms of time and other inputs per unit than producing just one or a few of many different things. This is called learning by doing. Because of learning by doing and other advantages of large scale production people do not typically produce the full range of goods and services on which they live. Instead we specialize, some producing one good, others producing other goods, some working as welders others as mothers, teachers or farmers. There are huge advantages to this pattern of specialization – called the division of labor. Those who are naturally better at some task, or have learned to be good at it by experience, or are in an environment in which it can be most productively done can devote themselves entirely to what they are relatively good at. This is part of the explanation of why as a species we are so productive. The limited number of species that have adopted a highly developed division of labor – humans, ants and other social insects, for example – have out competed other species. The total biomass of humans and the livestock we have domesticated, for example, is estimated to be 23 times the weight of all the other mammals on earth. And throughout most of human history the biomass of ants – one of the most cooperative of species – has exceed that of humans by a considerable amount. H I S TO RY The Israeli historian Yuval Noah Harari explains why it is our capacity to cooperate in flexible ways with large numbers of other humans that makes us unique among all the animals. https://tinyurl.com/y3bpy4px 294 MICROECONOMICS - DRAFT But the division of labor poses a problem for society: once they are produced by specialized labor, how are the goods and services to be distributed from the producer to the final user. In the course of history this has happened in a number of distinct ways from direct government requisitioning and distribution as was done in the U.S. and many economies during the Second World War, to gifts and voluntary sharing as we do in families today and was practiced among even unrelated members of a community by our hunting and gathering ancestors. In a modern capitalist economy, the institutions that govern how the goods and services are distributed from producer to user include markets, firms, families and governments. In this second part of our book we study markets and the actors who make up markets: the owners (and managers) of firms, and other individuals (and the families of which they are a part). To understand how markets facilitate specialization in chapter 6 we study the production process and how the division of labor and the exchange of products can be advantageous to all concerned. Then to understand the workings of markets we explain how individuals’ valuation of goods and services is expressed in market demands (Chapter 7). Then, along with these market demands, we explain how firms’ costs of production are expressed in their owners’ and managers’ decisions about how much to produce and supply to the market (Chapter 8). We then study the process of competition among sellers and buyers, each seeking to enlarge their share of the mutual gains made possible from the division of labor and exchange. And we show how this so called rent-seeking process affects the movement of prices and the quantities produced (Chapter 9). Taking these four chapters as a whole poses a tension that can be expressed by the following contradiction: • The models and evidence on the advantages of large scale production provide a reason why we specialize. • Competitive markets are essential to the process of specialization can be organized in ways that allow the mutual benefits of the division of labor to be widely shared, as Voltaire said "for the utility of mankind." • But the advantages of large scale production can also promote the emergence of giant firms and a winner take all process that appears to be making markets less competitive. Making market competition sustainable given the advantages of large scale production will have to be addressed by public policy. 6 Production: Technology and Specialization DOING ECONOMICS The division of labor is limited by the extent of the market. Adam Smith The Wealth of Nations A technician glances quickly from one to the other of her three monitors and around the huge room many other technicians do the same. Occasionally a technician looks up at gigantic blue video screens on which news reports flash, international weather reports display, and flight conditions stream live. Twenty-four hours a day, translators stand ready to facilitate conversations in 28 languages. What is this command center? It’s the Production Integration Center that coordinates the global production of the Boeing 787 Dreamliner four stories above the production floor at the company’s plant in Everett, Washington, USA. There the super-jumbo airplanes are being assembled from components being flown in from around the world: parts of the wing from Japan, wing tips from Korea, the center fuselage and the horizontal stabilizer from Italy, passenger doors from France, cargo doors from Sweden, landing gear from the U.K., and the list goes on and on. Figure 6.2 shows where the components of the Dreamliner are produced. In 2015, Boeing contracted with over twenty-six thousand suppliers around the world.1 Boeing selected Rolls Royce, Mitsubishi, Saab, Fuji and other companies to design and build the components because they were – in Boeing’s estimation – simply the best companies to do the job, anywhere in the world. The Japanese companies were global leaders in aircraft construction. The Italian partner Alena had critical intellectual property rights (patents) that Boeing would otherwise not have had access to. This chapter will enable you to do the following: • Explain how learning-by-doing and economies of scale are reasons for the division of labor and specialization. • Understand how markets allow specialization according to the principle of comparative advantage. • See how in the presence of economies of scale and learning by doing an economy can benefit by specializing, but also may specialize in ways that perpetuate is low income as a result of a poverty trap similar to coordination failures studied earlier. • Manipulate some commonly used production functions to study marginal and average products of the labor and capital goods and derive a production possibilities frontier. • Describe the main dimensions on which technologies differ – the extent of substitution among inputs, productivity, factor intensity and economies of scale – and how these are represented in different production functions. • Understand how a the owners of a firm can determine a set of inputs and a way of combining them to produce output (a technique of production) that will minimize the costs of a given level of output. • See that owners of a firm will try to innovate to reduce the inputs required to produce a given output and therefore low costs and receive innovation rents (at least until the competition catches up). Boeing modified four 747-400 aircraft – renaming them "Dreamlifters" – to deliver the wings, body, and other parts of the plane to Everett where American Figure 6.1: Boeing’s Production Integration Center in Everett, Washington, USA. 296 MICROECONOMICS - DRAFT Figure 6.2: The different parts of the Boeing 787 aircraft built by different specialized partners from all around the world. machinists and others assembled the planes. Four stories above them the engineers at the Production Integration Center kept minute-by-minute track of the movement of the components around the world. 6.1 The division of labor, specialization and the market Boeing’s globally integrated Dreamliner production process illustrates an important economic idea: the division of labor. The division of labor is an expression for the fact that people, organizations, or geographical regions specialize in particular tasks or the production of a limited range of goods or services. For Boeing, purchasing components of the Dreamliner from hundreds of other specialized firms was more cost effective than producing the entire plane in-house at their plant in Everettt, WA. There are two consequences of specialization. The first is increased productivity. The specialization allowed by the division of labor increases productivity for three reasons: • Comparative advantage: specialization enables people, firms, and regions to focus on the tasks and products that they are comparatively good at (we will take up comparative advantage below). • Learning by doing: People learn better ways of working both through the developing individual skills and discovering better ways to organize production among members of a team. Figure 8.2 b provides a dramatic example of learning by doing. • Economies of scale: By allowing the production of a few things on a large scale rather than many things on a small scale, specialization raises the amount of output that is available for a given amount of inputs. Figure 8.6 presents some physical evidence for economies of scale based on engineering studies. A second consequence of specialization is the need for integration. The P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N advantages of the division of labor can only be realized if there are institutions to coordinate the many distinct production activities that take place when people specialize. The Boeing example illustrates the need for integration: somebody has to put the parts together to produce a Dreamliner. This can be summarized: production is specialized, but the use of goods and services is generalized. Specializing in consumption is not biologically sustainable. As a result, the goods and services, somehow have to get from specialist producers to generalist users. To grasp the scope of this problem imagine a 3D map of the world showing the stocks of goods produced annually in each location. There would be a hundred or so Dreamliners piled up in Everett Washington, billions of square meters of cloth stacked in Bangladesh and other textile producing countries, well over a billion barrels of oil piled over tiny Kuwait, mountains of computer components and other consumer electronics rising from coastal China were Dell is located, and so on. Now imagine the same map, but showing where all of the goods are used. The second map would be different from the first in two ways: • it would be much flatter, the goods would have been spread around to the entire population of the world and • in any location there would be an assortment of a great many products, not towering stacks of a single product. The coordination of specialized producers and generalist users is accomplished by a set of institutions that differ in importance both over time and across the economies of the world today. These include: • Market exchanges: Selling the goods that specialized producers have made provides the budget for purchasing the general market basket of goods and services on which we live. • Government acquisition and provision: Publicly provided services are based on the integration of the specialized producers goods and services to provide education, security and other government-provided services to generalist users of these services. • Families and other face-to-face communities: Typically families exhibit a division of labor by age and gender: adult women, for example, biologically producing children and spending disproportionate time on raising them and caring for other family members (for example preparing meals). The goods produced and tasks performed by adult men and women and by children are shared within the family or some other larger consumption unit. These three ways of coordinating the division of labor have in common that they are like a two-sided platform that connects specialist producers with gen- 297 298 MICROECONOMICS - DRAFT 0.72 Slope of ray from origin equals Average Product Total product, x Total product, x x= Total Product 1 x = (l)2 50 0.97 Slope of tangent line equals Marginal Product 0.55 4 2 6 Marginal product 1 mp(l) = (l) 25 0.24 0.16 0.12 Average product 1 ap(l) = (l) 50 0.08 0.04 2 4 Average & marginal product, ap, mp Average & marginal product, ap, mp Slope of ray from origin equals Average Product Slope of tangent line equals Marginal Product 2 (l)2 Total product 1 x = ln(1 + l) 2 0.8 0.32 0.08 1 50 4 6 0.27 Average product 1 (ln(l + 1)) 2l 0.2 0.17 0.16 ap(l) = Marginal product 1 mp(l) = 2(l + 1) 0.1 0.07 6 Hours of labor, l (a) Economies of scale eralist users of goods and services. – like Air BnB that matches home owners to people looking for a place to stay, or Tinder, a dating app. This chapter’s head quote by Adam Smith tells us that markets play a critical role in allowing the division of labor to expand to global proportions, leading to ever greater specialization. Here, in Part III of this book and also Chapter 14 we explain how markets work to coordinate the division of labor. We begin with the production process, and some aspects of it that favor specialization. 6.2 Production functions with a single input In Chapter 3, we used information on the way that Aisha’s study time translated into her learning to ask how much time she will choose to study. In Chapter 5 information about the relationship between fishing time and the amount of fish caught was a key idea in posing and then addressing the common property resource coordination problem. In both cases we were using, as you recall from Chapter 5, production functions. To better understand the division of labor and specialization we now need to look more carefully at the properties of production functions. Think about another person, Alex, who has to choose how much of his time to spend produc- 2 4 6 Hours of labor, l (b) Diseconomies of scale Figure 6.3: Production functions with economies and diseconomies of scale. With economies of scale (Figure a) doubling labor input more than doubles output, as can be seen by going from 2 hours of labor to 4 hours. Average product and marginal product increase with labor input as you can see from the slope of the production function (the marginal product), which steepens as the labor input increases and the slope of the ray from the origin to a point on the production function (the average product), which also steepens. The average and marginal products given by these slopes in the upper figure are shown in the lower figure of panel a. With diseconomies of scale (panel b), the opposite occurs: the average and marginal products of labor are both decreasing. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 299 ing one or more goods. Alex can devote more or less labor (l ) to production and he can observe how his output (x) varies as he changes the number of hours he works. Alex may use a computer and a desk or a given plot of land and farming equipment or other inputs, but for now we assume that the only input is his labor time. The relationship between the input of his labor and the output of the goods is described in a production function – x = f (l ) – taking the form: x(l ) = ql a (6.1) The exponent a measures the responsiveness of output to a change in the R E M I N D E R A production function is a mathematical description of the relationship between the quantity of inputs devoted to production on the one hand (the arguments of the function) and the maximum quantity of output. level of labor. The positive constant q measures the overall productivity of the production process, which will be greater the more skilled or hard-working Alex is. The top panel in Figure 6.3 a illustrates this production function (equa1 tion 6.1) with q = 10 and a = 2. The top panel in 6.3 b shows a different production function: x = 12 ln(1 + l ). In the top portion of both panels more hours of labor result in more output, but the panels differ in how much output is obtained for given inputs. E CONOMIES AND DISECONOMIES OF SCALE When production exhibits economies of scale, increasing inputs by a factor more than proportionally increases output; with diseconomies of scale, increasing inputs by a factor less than proportionally increases output. • Economies of scale (6.3 a): when Alex doubles all of the inputs – in this case that means just his labor input – the output more than doubles. • Diseconomies of scale (6.3 b): when Alex doubles his labor input (assumed to be the only input), the output less than doubles. The term constant returns to scale, not shown in the figure, refers to the case where when inputs double output doubles, so the production function is just a straight line as shown in the lower right panel of Figure 6.4. In the lower figures of both panels we show two important statistics describing aspects of the two production functions in the above figures. The ratio of the amount of output to the amount of the input involved in producing it is the average product of that input (also called average productivity). Average product is measured by the slope of the line from the origin (called a AVERAGE PRODUCT The average product of labor is the ratio of the output to the labor input. ray) to a point on the production function. The bottom panel in figures a and b show the average product of labor associated with the production function shown at the top of those figures. When production exhibits economies of scale, average product increases as the scale of production increases through an increase in inputs. The ratio of the increase of output to an increase in labor input is the marginal product of labor (also called marginal productivity). M-Note 6.1 summarizes the different cases when the output is x and the only input is labor, l . M ARGINAL PRODUCT The marginal product of labor is the ratio of the change in total output to a small change in input. 300 MICROECONOMICS - DRAFT Checkpoint 6.1: Production and labor inputs Consider a production function: x(l ) = 10l 0.5 : a. Sketch the production function. b. Calculate ap(l ) and mp(l ). Sketch them. c. Does the production function exhibit economies or diseconomies of scale? M-Note 6.1: The average and marginal product Here we summarize the concepts of total product, average product and marginal product. Because the marginal product is the slope of the production function, it is also the derivative of the production function with respect to a particular input, e.g. for a production function using only labor, xl = d f (l ) dl . If there is just a single input, labor, and the marginal product of labor is greater than the average product of labor then the average product must be increasing as more labor is used. We will use equation 6.1 to illustrate why this must be the case. We start calculating the average product ap and marginal product mp of the production function x(l ) = ql a : Average product: ap = Marginal product: mp = x(l ) ql a = = ql a l l dx(l ) = aql a 1 dl 1 (6.2) (6.3) To analyze how the average product changes as more labor is put into production, we calculate the derivative of the ap function with respect to labor hours: dap(l ) dl = (a 1)ql a 2 If a > 1, the expression above is positive, which means that the ap increases with more labor. If a < 1, it is negative: the ap is reduced if we add labor. In summary: • If a > 1, then mp > ap, and so • If a < 1, then mp < ap dap(l ) dl dap(l ) and so dl > 0 and <0 6.3 Economies of scale and the feasible production set Suppose that Alex can spend his time fishing and making shirts in some combination, including complete specialization (spending all of his time on one or the other). He prefers to have more of both shirts and fish: both are goods. He needs at least some of each to survive. His labor time, as in Chapter 5 is a "bad" but we will set aside his choice of total hours of work by saying that he can work any amount up to ten hours a day, and that given how productive his labor is and how much he values the goods, he will choose to work the full 10 hours. As a result, the more time Alex devotes to producing one good, the more of that good he will have, but because Alex’s time is limited, the less he can produce of the other. Therefore the opportunity cost of more fish P RODUCTION POSSIBILITIES FRONTIER (PPF) The production possibilities frontier for two goods shows the maximum feasible amount of one good that can be produced given the output of the other. The production possibilities frontier is the boundary of the producer’s feasible set and is an alternative name for the feasible frontier when we study on production. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 301 is the amount of shirts he will have to forego if he shifts his work time from shirt-making to fishing. In order to pose the question – how much time will he spend on each? – we need two pieces of information: • the feasible set of combinations of fish and shirt amounts that are available to him, given his labor time and the production functions at his disposal; and • the indifference map representing his valuation of each of the combinations of the two goods. We derive the feasible set in this section and introduce the indifference map in the next. We will assume that there are economies of scale in shirt making (as is common in manufacturing processes) but constant returns to scale in fishing. This means that the fish production function is linear: it is a straight line. As a result both the average and marginal products of labor are constant and equal to each other and also equal to the slope of the blue line through points f an c in Figure 6.4. In in Figure 6.4 we derive his feasible set of fish and shirts based on • the total amount of time he will work (shown in the lower left quadrant of the figure); and • the productions functions for fish and shirts (shown in the lower right and upper left quadrants, respectively). In the figure the horizontal axis to the left of the origin represents positive amounts of labor devoted to making shirts, and the vertical axis below the origin is positive amounts of labor devoted to fishing. Alex needs to make a choice between three different ways to allocate his time in production to two types of output (fish and shirts): a. Allocate 10 hours of work to producing only shirts b. Allocate some hours of work to producing shirts and some to fish, and c. Allocate 10 hours of work to producing only fish Option a) is shown as point a in the figure where Alex dedicates all of his 10 hours of work time to producing shirts. We extend a line up to his production function for shirts and notice that 10 hours of labor results in Alex producing 50 shirts. We extend a dashed line to the y-axis to see what this would correspond to on the production possibilities frontier and see that Alex would produce 50 shirts and no fish as a result of dedicating all his labor to shirts. We could follow the M - C H E C K Remember, the average product is the slope of a ray from the origin to a point on the production function, and the marginal product is the slope of the production function. So if the production function is just a straight line, both of these are equal and do not vary as more labor is devoted to production. MICROECONOMICS - DRAFT a 10 10 hrs of labor for shirts produces 10 shirts Shirts, y 302 Shirt Production 1 y = (ls)2 10 e 10 2.5 Feasible set of outputs Labor for Shirts, ls 5 b 5 2.5 Fish, x Figure 6.4: Deriving the production possibilities frontier with economies of scale. The lower left quadrant shows the constraint: a given amount of labor is available. The upper left and lower right show how the available labor can produce shirts and fish respectively. The upper left is an economies of scale production function similar to the panel a in Figure 6.3, but just rotated clockwise 280 degrees. The lower right production function has constant returns to scale. Points d, e, f, and b illustrate the production of shirts and fish that are possible if the labor time is divided equally among the two sectors. To check that you understand how the figure works, find the point on the feasible frontier associated with devoting 8 hours to fishing and 2 to shirt-making, (trace out the new points, d’, e’, f’, and b’). Feasible set of labor hours d 5 Constraint on total labor hours S l + lF ≤ 10 10 Labor for Fish, lf lS = 5, lF = 5 f Fish Production 1 x = (lf) 2 c 10 hrs of labor for fishing produces 5 kgs of fish same process for option c) corresponding to point c in the figure. He would produce 5 fish and no shirts by dedicating all his time to fishing. Option b) (that is, point b) on the other hand, shows what Alex would produce by dedicating half his time to shirts and half to fishing. Because production in fishing is linear, if he dedicates 5 hours, he simply gets half of what he would have produced at 10 hours (2.5 fish). But, because there are economies of scale in shirt production, from dedicating half his time to shirts, he only gets a quarter of the output relative to 10 hours of labor for shirts (12.5 shirts vs. 50 shirts) The top-right quadrant of Figure 6.4 illustrates economies of scale. The result is that Alex’s production possibilities frontier is bowed inward toward the origin. This reflects the fact that with economies of scale in one or both production functions, dividing your work time between the production of both is not as good (it is closer to the origin) than devoting all your time to just one or the other. The (negative of the) slope of the production possibilities frontier shows the opportunity cost of acquiring more fish by shifting labor from shirt-making to fishing, in terms of the amount of shirts that must be foregone as a result. With economies of scale, as Alex shifts his labor input from producing clothing to producing fish, mrt (x, y) declines, so he gives up smaller and smaller amounts of clothing to get larger and larger amounts of food. This reflects the fact that with economies of scale the marginal product of M AT H N OT E We describe the production possibilities frontier with economies of scale as convex toward the origin, that is, it is bowed in to the origin. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 303 his labor decreases the less labor he devotes to production of a good. This means that when he is doing little shirt-making, his marginal productivity in that activity is low, so doing a little less (so as to allow him to do more fishing) does not result in a large reduction in shirts produced. 6.4 Economies of scale, specialization and exchange In Figure 6.5 we combine the feasible set from the Figure 6.4 with an indifference map represented by the three numbered indifference curves. Recall that, as indicated by the numbering of the curves, farther away from the origin is better in Alex’s evaluation of outcomes because both shirts and fish are goods. Diversification in the absence of exchange If Alex cannot exchange the goods with others, does the best he can by following the mrs = mrt rule and finding the point on the production possibilities frontier that is tangent to the highest indifference curve, at point d (for diversified production), and consuming xd and yd . In this case Alex chooses to produce some of both goods because the marginal utility of each of them is diminishing the more he consumes, so having some of both is superior to having all of one kind. Remember this is why his indifference curves are bowed inward toward the origin. "The division of labor is limited by the extent of the market" But, if the producer can exchange the goods with others, then there is a second way that he can "transform shirts into fish." He does not do so by reallocating his time from shirt making to fishing. Instead he can spend all of his time making shirts and then exchange some shirts for some fish if he can find a willing buyer for his shirts. Suppose such a trader is found, and she is willing to buy any amount of his shirts at a given price ( p): in return for p shirts she is willing to provide 1 kg of fish. This is the second way of transforming shirts into fish, and the marginal rate of transformation is p: the quantity of shirts that one has to give us in exchange for a kg of fish. This opportunity for exchange alters the feasible set constraining what Alex can do, as shown in Figure 6.5. The orange line with the y-intercept at ȳ (which is the maximum amount of shirts Alex can produce) represents his new feasible frontier with exchange. Its slope is -p, the (negative of the) opportunity cost of acquiring more fish in terms of the shirts foregone. This means that marginal rate of transformation of shirts into kg of fish is just p: giving up p shirts gets you 1 kg of fish. Here the process depicted by movements along R E M I N D E R Remember that production functions with economies of scale are convex (output increases at an increasing rate with input) and production functions with diseconomies of scale are concave (output increases at a decreasing rate with input). 304 MICROECONOMICS - DRAFT y = 10 y = 10 = ys s yd = 4.5 Quantity of shirts, y Quantity of shirts, y Feasible frontier when specializing in shirts mrs(x, y) = mrt(x, y) d e ye = 4.9 yd = 4.5 mrs(x, y) = mrt(x, y) d uA3 uA2 Feasible outputs 0 xd = 1.65 xs = 0 x=5 uA1 xd = 1.65 Kilograms of fish, x xe = 3.4 x=5 6.67 Kilograms of fish, x (a) Economies of scale without trade (b) Economies of scale and trade the price line shows exchange in varying amounts, not shifting Alex’s own labor from making shirt to fishing. Because at the price p, 1 kg of fish is worth the same as p shirts, then the value of all of the combinations of quantities of fish and shirts along the orange price line in Figure 6.5 b. have the same value. This is because the value (expressed in number of shirts) of the fish purchased p · x must be equal to the value of the shirts sold (which is just the number of shirts sold, ȳ p · x = ȳ uA2 Feasible outputs uA1 y) or: y (6.4) The constrained optimization problem that Alex faces comes in two steps: • Step 1: Decide on whether to specialize and if so, in which good; to do this find the distribution of labor time between fishing and shirt making that maximizes the value of one’s output then • Step 2: Decide whether to exchange any of the goods produced, and if so how many; to do this maximize utility subject the new feasible frontier given by the goods produced and the relative price. In Figure 6.5, Alex will decide to produce at point s (for specialized production) at the intercept of his production possibilities frontier with the shirt axis to maximize the value of his output at the relative price p. Then he will exchange the shirts he produces for fish at the price p to reach point e (for exchange) on the highest indifference curve in his new feasible set, uA 3 . This is where: marginal rate of substitution = p = marginal rate of transformation by exchange Figure 6.5: Production possibilities frontier (PPF) with economies of scale. In Figure a, we present the producer’s choice when they do not have the opportunity to trade. Using the mrs = mrt rule, he producer chooses the point at which their indifference curve is tangent to their production possibilities frontier at point d. In Figure b, we show what happens if the producer can exchange shirts for fish at some constant price ratio. In this case, he can do better by specializing in shirt production (good y) and then acquiring the fish she desires through exchange, not by producing them. The green shaded area is the enlargement of his set of feasible levels of consumption of the two goods. He produces shirts only at point s, then exchanges the shirts on the market for fish, taking him to his higher indifference curve, u3 at point e. Notice that in selecting point s the producer is not implementing the mrs = mrt rule. The reason is that he does better at the corner solution, producing none of the x good at all. But in choosing point e by exchange, he is consuming both goods (not a corner solution) and so the mrs = mrt rule implements his constrained utility maximum. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N Shirt production Figure 6.6: Production possibilities frontier (PPF) with diseconomies of scale. The fourquadrant graph shows a production possibilities frontier with diseconomies of scale in production in the top-right quadrant. The diseconomies of scale depicted in the production possibilities frontier arise from a relationship in the production technologies from the two different sectors: fishing (bottom right quadrant) and shirt-making (top-left quadrant). A worker is constrained by how much of their labor time they can dedicate to either producing fish or producing shirts. Their labor constraint is depicted in the bottom-left quadrant, shown as a limit of 10 hours of labor per day. Shirts,y 10 hrs of labor for shirts produces 8 shirts Feasible frontier 5.66 1 y = 2.53(ls)2 Feasible outputs Labor for shirts, ls 10 5 5.66 Feasible labor hours ls = 5, lf = 5 305 Kgs of fish, x Fish production 1 x = 2.53(lf)2 ls + lf ≤ 10 Constraint on total 10 labor hours Labor for fish, l f 5 10 hrs of labor for fishing produces 8 kgs of fish Our example – Alex choosing what to produce – demonstrates two general truths: • If one or more production function with economies of scale is available, it may make sense to specialize but • this will be true only if there are others producing different goods and there are opportunities for exchange, integrating specialized producers with generalist users to coordinate the division of labor. This is the basis of the interdependence of different producers within the division of labor. Checkpoint 6.2: The choice of what to specialize in E X A M P L E In modern economies a household may specialize in providing labor with some particular mix of skills, training and experience to an employer. • Redraw Figure 6.5 with a higher relative price of fish (so p, the number of shirts that one must give up to get a kg of fish is now larger). • Show that if p is sufficiently high, Alex will do better specializing in fish (indicate the amount he will produce, and the amount he will exchange). Diseconomies of scale, diversification and exchange In contrast with the production with economies of scale illustrated in Figure 6.4, Figure 6.6 illustrates the case of diseconomies of scale, in both fishing and shirt making. The result is that Alex’s production possibilities frontier is bowed outward from, or concave to, the origin. With diseconomies of scale, as Alex shifts his labor input from producing shirts to producing fish, he gives M - C H E C K The production possibilities frontier with diseconomies of scale in both production functions is concave toward the origin, or bowed in. 306 MICROECONOMICS - DRAFT e ye = 9 Quantity of shirts, y y=8 mrs(x, y) = mrt(x, y) Feasible frontier Price line (feasible frontier) yd = 5.7 d s ys = 3.6 uA3 uA2 uA1 xe = 4.5 xd = 5.7 xs = 7.2 x = 8 Kilograms of fish, x up larger and larger amounts of shirts to get smaller and smaller amounts of fish. This reflects the fact that with diseconomies of scale the marginal product of his labor decreases the more labor he devotes to production of a good. With diseconomies of scale the mrt (x, y) increases as labor is reallocated from shirts to fish reflecting the idea of increasing opportunity costs. If he did not have opportunities for trading goods, he would select point d in Figure 6.7 with utility uA 2. But he can do better if he decides what to produce knowing in advance that he be able to exchange the goods he produces. So he will use the two-step constrained optimization procedure outlined above: first decide what to produce so as to maximize the value of his output, then exchange goods to maximize his utility. But due to the diseconomies of scale the result is not complete specialization. He will decide to diversify. He will produce at point s somewhere in middle of his production possibilities frontier where his mrt (x, y) is equal to the relative price p, putting labor into producing both goods, to maximize the value of his output at the relative price p. But the possibility of exchange expands his feasible set: the orange line is the feasible frontier with production at point s and exchange possible at price p. He will then exchange one or the other of the goods he produces for the other – in this case exchanging fish for shirts – to reach the highest indifference curve that is in his feasible set, uA 3. Figure 6.7: The production possibilities frontier for fish and shirts when there are diseconomies of scale in production. If the producer cannot exchange the goods with others, he does the best he can by finding the point on the production possibilities frontier that is tangent to the highest indifference curve, at point d, and consuming xd and yd . But, if the producer can exchange the goods with others, the producer chooses the production point with the highest value at that price, and then exchanges output to maximize utility at point s and then moves along the price line to point e on u3 . P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 307 6.5 Comparative and absolute advantage The question ”What should you specialize in?” seems to have an obvious answer: ”Specialize in what you are best at.” The same would seem to go for countries: they should specialize in what they are best at producing. But what, exactly, does that mean? ”Better” than other people (o countries)? What if you are not better than others at anything? Should you not specialize in anything? Differing opportunity costs and comparative advantage Or, does “better” mean ”better than you are at other things that you could do”? If that’s what it means – at least compared to how good others are in those same things – then we are talking about comparative advantage. To see what this means, suppose that a recent graduate, Brett, has started a data science business. When Brett writes reports for his business, there are two tasks: entering data in some digital format and generating graphs to some detailed specifications using the digitized data. Lets say that each graph requires 1,250 keystrokes of data. Brett has the option of doing both tasks, or doing either one of them himself and getting the other done for pay on Mechanical Turk (which calls itself ”the online marketplace for work”). There are many people like Brett, some of them offering their services on M-turk, as it is called, and with their pay purchasing other services from M-turk. But we will consider just one of these people named April (there are lots of people like her too). When we say April, we really mean "people like April ready to sell graphs on M-turk, in return for data entry sold by people like Brett." And a similar statement goes for Brett. The reason is that if we had some particular Brett exchanging with a particular April, then in agreeing on a price they would have to agree on the amounts to be exchanged (as Ayanda and Biko did in Chapter 4). This would be an unnecessary complication, so we avoid it by assuming that both Brett and April can purchase and sell as much as they like at whatever price is posted on M-turk. M-Note 6.2: Opportunity costs, feasible frontiers and comparative advantage To understand how production functions, opportunity costs, and the feasible frontier determine absolute and comparative advantage we use the following notation: 1 • ax = time (fraction of an hour) required to input 1000 keystrokes of data ( 11 for April 1 and 10 for Brett, so April has an absolute advantage in data entry.) 1 • ay = time (fraction of an hour) required to produce one graph ( 20 for April and 18 for Brett, so April has an absolute advantage in graph-making) O NLINE M ARKETPLACESMechanical Turk is one of many online marketplaces for work tasks for pay. Others would include Clickworker, Fiverr, UpWork and many others. People can be paid be for small, short tasks like data input (which is more typical for sites like M-Turk and Clickworker) or for more advanced jobs like Fiverr and UpWork. 308 MICROECONOMICS Graphs made, y 20 - DRAFT Figure 6.8: Feasible frontiers: absolute and comparative advantage. April has an absolute advantage in the production of both goods because her feasible frontier is outside Brett’s. Brett has comparative advantage in data entry because his feasible frontier is flatter than hers (lower opportunity cost of data entry). Without the possibility of exchange Bret completes 4 graphs (at point g). Remember: each graph requires 1250 keystrokes of data (also remember that the horizontal axis of Figure 6.8 is measured in thousands of keystrokes). This means that they must be on the dashed orange line from the origin. The question is how far out they can get. This x dashed line for y = 1.25 lets us see how many complete graphs (data entry and graph preparation combined) each person could be by themselves in one hour. ffA Each graph needs 1,250 keystrokes x y= 1.25 8 ffB i 6.11 g 4 5 10 11 7.64 Data entered ('000's), x • x = thousands of keystrokes of data entered • y = number of graphs made • T = total time = 1 hour Total labor time is composed of time spent entering data plus time spent producing graphs, so the feasible set is defined by: Time constraint ax x + ay y T (6.5) The equation for the feasible frontier is Equation 6.5 expressed as an equality and rearranged with y as a function of x. That is, the feasible frontier could be depicted as: Feasible frontier y = T ay ax x ay (6.6) Which means that: dy ax = = mrt dx ay (6.7) dy dx is the negative of the slope of the feasible frontier (which can be seen from Equation 6.6). This is also the opportunity cost to either person for producing data entry in terms of graphs. But we can also use the numbers to find their opportunity costs. For April: mrt (x, y) = dy ax = = dx ay mrt (x, y) = dy ax = = dx ay 1 11 1 20 = 20 = 1.82 11 (6.8) = 8 = 0.8 10 (6.9) For Brett: 1 10 1 8 Because 0.8 < 1.82, Brett’s comparative advantage is in data entry. Figure 6.8 shows how good April and Brett are at the two tasks, as indicated P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 309 by their feasible frontiers if they are restricted to working for just one hour. The points making up their respective frontiers are the combinations of outputs from data entry and graph making that use up one hour of their working time. For example, in an hour Brett can produce 8 graphs and enter no data, or 10 (thousand) keystrokes of data entry, and no graphs, or 4 graphs and 5 (thousand) keystrokes, and so on. Figure 6.8 illustrates the concepts of both absolute and comparative advantage. April has an absolute advantage in producing both goods: data entry in thousands (x) and graphs (y). A person has an absolute advantage in the production of a particular good if, given the set of available inputs, she can produce more of it than some other person. M-note 6.2 shows how the feasible frontier is derived from the data on the productivity of April and Brett at the two tasks. In our case, absolute advantage means that, in one hour, if April devoted all of her time to data entry, she could enter more thousands of keystrokes of data than Brett (11 rather than 10) and likewise for making graphs (20 rather than 8). As her feasible set includes Brett’s entire feasible set (her feasible frontier is farther from the origin), she can produce more in an hour than Brett can in any combination – complete specialization in one or the other or some ratio of data entry to figure making. Different opportunity costs: The basis of specialization and exchange This raises the question: if April is better at both data entry and graph making, why would she want to trade with Brett at all? This is where the concept of comparative advantage comes in. A person has a comparative advantage in the production of a particular good if their opportunity cost of producing that A BSOLUTE ADVANTAGE A person has an absolute advantage in the production of a particular good if, given the set of available inputs, she can produce more of it than another person. C OMPARATIVE ADVANTAGE A person has a comparative advantage in the production of a particular good if the opportunity cost to them of producing that is lower than it is for another person. good is lower than it is for another person. For Brett, spending the hour it would require to enter 10 thousand more keystrokes of data would mean that he could not make 8 graphs. So 8 graphs is his opportunity cost of 10 thousand keystrokes of data entry. Translating this to be in the units of the figure, 0.8 graphs is the opportunity cost to Brett of 1 thousand keystrokes of data entry (which is the negative of the slope of his Maximum possible data entry Brett April 10 11 8 20 0.8 1.82 1.25 0.55 (thousands of keystrokes per hr) Maximum possible graphs (graphs per hr) Opportunity cost of 1 thousand keystrokes of data entry (in graphs) Opportunity cost of making 1 graph (in thousands of keystrokes) Table 6.1: Absolute and comparative advantage: Number of bits of data and graphs created in one hours work. The entries in blue show that April has the absolute advantage in producing both data entry and making graphs. The entries in red show that Brett has a comparative advantage in producing data entry (0.8 < 1.82) and similarly April has a comparative advantage in making graphs (0.55 < 1.25). Remember, if they are working alone, they would never produce only graphs or only data, because they need a combination of graphs and data for the project. Additionally, note that, for both of them the two opportunity costs are simply the 1 inverse of one another, e.g. for Brett 1.25 = 0.8 . 310 MICROECONOMICS - DRAFT feasible frontier). By contrast, for April, 10 thousand keystrokes of data entry requires just 55 minutes (she enters 11 thousand keystrokes per hour) and in that period of time she could have made 18.2 graphs. So for April the opportunity cost of 10 thousand keystrokes is 18.2 graphs, or translating this to the quantities in the figure, the opportunity cost of a thousands keystrokes is 1.82 graphs. Brett’s comparative advantage is in data entry. This is not because he is so good at data entry; April is better at data entry then him. It is because he is so unproductive in producing graphs, so the opportunity cost of taking time away from graph-making to do data entry (the graphs he otherwise could have made) is low. It can similarly be seen that April’s comparative advantage is in producing graphs. Table 6.1 summarizes Brett and April’s absolute and comparative advantage in these tasks. Here is a simple way to remember the difference between absolute and comparative advantage: • If for a given axis (horizontal or vertical) the intercept of one person’s feasible frontier is outside (farther from the origin than) the other’s, then that person has an absolute advantage in the good on that axis. • Comparative advantage is determined by the slope of the feasible frontier : The comparative advantage of person with the flatter feasible frontier is in the good on the horizontal axis. This is because the (negative of the) slope of the feasible frontier (how steep it is) is the opportunity cost of the good on the x axis. The comparative advantage of the person with the steeper feasible frontier is in the good on the vertical axis. The second bullet says that unless the two feasible frontiers have the same slope, the comparative advantage of the two people will differ. Even though one of them may not have an absolute advantage in either good (like Brett) each will have a comparative advantage in one of the goods. So Brett has to be comparatively good at something. The data show that the opportunity cost of data entry is less for Brett than it is for April. We now show why this provides the basis for Brett specializing in data entry, April spending all her time making graphs, and the two entering into an exchange. Checkpoint 6.3: Comparative and absolute advantage a. If April could make only 7 graphs in an hour (Brett’s productivity remaining unchanged) which of them would have an absolute advantage in which of the goods? In which good would April’s comparative advantage be? b. If in an hour Brett could enter only 4,400 keystrokes of data, who would have P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N comparative advantage in data entry? 6.6 Specialization according to comparative advantage In Figure 6.8 we can see that if Brett produced both data entry and graphs himself, he would get to point g – four completed graphs based on the required 5 thousand keystrokes of data. Similarly, April working by herself could produce 6.11 graphs along with the necessary data. Will the two be able to do better by specializing in the production of just a single good each, and then exchanging graphs for data entry, so that each would have the required keystrokes of data for each graph? The opportunity to exchange expands the feasible set To see that they will, think about two ways that April can get data entered: she can do this herself, with every thousand keystrokes entered bearing an opportunity cost of 1.82 graphs not made. Or she could pay someone else to enter data, paying with some of her graphs. What is the most she would be willing to pay for a thousand keystrokes of data entered? The answer is 1.82 graphs which is what she would have to "pay" in graph-making foregone, if she did the data entry herself. This is her maximum willingness to pay for data entry. This is also the slope of her feasible frontier. Would Brett be willing to sell her data entry for a price less than 1.82 graphs per thousand key strokes? The lowest price at which he would sell a thousand keystrokes of data entry is 0.8 graphs because this is his opportunity cost of data entry. His opportunity cost is the number of graphs he gives up producing if he enters a thousand more key strokes. This price is called Brett’s minimum willingness to accept (giving up data in return for graphs). This is the slope of his feasible frontier. Because April’s willingness to pay is greater than Brett’s minimum willingness to accept – the lowest price at which he would sell keystrokes – each of them can benefit by specializing and then entering into an exchange. In Figure 6.9 we show what Brett can accomplish when he specializes in data entry and then exchanges some data entry for the graphs he needs to complete his project. The exchange opportunities are shown by the price lines (which are parallel because both people face the same relative prices). The (negative of the) slope of the price line is the number of graphs that can be purchased with a thousand keystrokes of data entry, or 1.45 in our example illustrated in the figure. Steeper is better for Brett, flatter is better for April. 311 312 MICROECONOMICS 20 - DRAFT Figure 6.9: Feasible frontiers and relative prices for exchange. Without the possibility of exchange Brett completes 4 graphs (at point g). The two arrows show that instead, he could move to point sB , specializing entirely in data entry, and then exchange some of his data entry with April in return for her making 5.16 graphs for him. The arrows at the top show how April could specialize and exchange. sA Graphs made, y 14.5 Each graph needs 1,250 keystrokes x y= 1.25 price line 8 7.11 j i ffB 5.16 h g ffA sB 6.45 8.88 10 11 price line 13.8 Data entered ('000's), x Specialization and mutually beneficial exchange To see why specialization and exchange will be mutually beneficial, you can think of the following: • The (negative of the) slope of the price line as as the marginal rate of transformation of keystrokes into graphs by means of exchange. • The (negative of the) slope of the feasible frontier is the marginal rate of transformation of of keystrokes into graphs by means of devoting more time to graph-making and less to data entry. The possibility of exchange gives Brett a new feasible set, with the frontier being the price line passing through any point on his "working alone" feasible frontier, indicating his exchange opportunities when he can buy 1.45 graphs with a thousand key strokes. With this new opportunity he could move in two steps from point g to point h. He could do this if he, first, specialized at point sB and then, second, engaged in exchange, moving up the price line to point h. Similarly, if April were at point i producing both goods, she could move to point j if she specialized at point sA and engaged in exchange to take her down her price line to point j. In Table ?? we compare their situation when producing both goods with the outcome when they specialize and trade. The reason why a mutually beneficial exchange would be possible is that the price at which the exchange took place (1.45 graphs per 1000 keystrokes) P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N Brett April Working independently Data entry 10,000 (1 hour) 15,280 (1.39 hours) Making graphs 8 (1 hour) 12.22 (0.61 hours) Graphs submitted 8 12.22 for the project Specializing and trading Data entry 20000 (2 hours) 0 Making graphs 0 40 (2 hours) Work produced 12,900 keystrokes 14.22 graphs 7,100 keystrokes 25.78 graphs Others’ work purchased 10.32 graphs 17,775 keystrokes Project submitted 10.32 graphs 14.22 graphs for own project Work for pay to exchange with others was greater than Brett’s opportunity cost of keystrokes and less than April’s opportunity cost of keystrokes. Or returning to Figure 6.8 the slope of the price line was greater than the slope of Brett’s feasible frontier and less than the slope of April’s. We have not explained why this particular price was the one at which they traded (as this would have been distraction from introducing comparative advantage). But any price between 0.8 and 1.82 would have allowed mutually beneficial exchange to take place. What made specialization possible in this case is two things: • Differences: Brett and April differed in their comparative advantage so there was some price – 1.45 per 1000 is just one example – at which they could both benefit form an exchange. • Opportunities for exchange: There was a way to exchange one’s completed tasks with others so as to obtain the right mix of data entry and graph making. Checkpoint 6.4: The distribution of the gains from specialization and exchange a. Using Figure 6.8, determine the price (graphs per thousand keystrokes) such that Brett would not benefit at all from specializing and trading and also the price ratio such that April would not benefit. b. Use the required number of thousands of keystrokes per graph (1.25) to say how many graphs Brett could make if the price at which he could sell 1000 keystrokes fell from 1.45 graphs to 1 graph. 313 Table 6.2: Specialization and exchange according to comparative advantage. The price of a thousand keystrokes of data entry is 1.45 graphs. Remember Brett is exchanging his data entry for graphs with people like April (not just April herself). And the same goes for April’s exchanges. This is why it is possible for the number of graphs that our particular Brett purchased (5.16) to differ from the number of graphs that April sold (12.89). 314 MICROECONOMICS - DRAFT 6.7 History, specialization, and coordination failures Brett and April simply decided to specialize in the tasks in which each had a comparative advantage. The existence of the online marketplace for tasks made this possible, and both people benefited by comparison to producing their reports without specializing. In the earlier example, Alex simply chose to produce shirts rather than fish, and he was able to feed himself because he could exchange shirts for fish. These personal examples have important lessons about specialization. But comparative advantage is more often applied to what countries do, not to what people do. When it comes to countries, we cannot say that, for example, Germany "decided" to specialize in machine tools and Bangladesh in textiles. What countries specialize in is the result of decisions made by vast numbers of people independently choosing what kinds of skills they will learn, the jobs they will take, what kinds of products the firms they own will produce and similar decisions. Countries – unlike Brett and April – can sometimes end up specializing in such a way that they remain poor. Had they specialized in something else, they would have been rich. To see how countries can specialize and stay poor, return to the feasible frontier in Figure 6.5. But now think about the figure as applying to an entire country, not just choices that Alex might make between fishing and making clothing. In this case, the economies of scale in the production of shirts occur because in every firm labor is more productive in producing shirts the more shirts are being produced in all of the clothing industry. Industry-wide (rather than firm-level) economies of scale are called economies of agglomeration. Economies of agglomeration means that the productivity of labor is greater the larger is the total output of the many firms producing similar goods in one country or region. Economies of agglomeration contribute to the geographical concentration of particular industries, for example: • software engineering in Bangalore (Bengaluru), India • finance in Hong Kong, London and New York City • information technology and IT related production in Silicon Valley, California • machine tools and motor vehicles in the Stuttgart-Munich region of Germany Economies of agglomeration occur because when large numbers of people are employed in producing the same product, the skills and other knowledge particular to that industry are widely diffused in the population, resulting in E CONOMIES OF AGGLOMERATION refers to cases in which the productivity of labor is greater, the larger is the total output of the many firms producing similar goods in one country or region. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N Quantity of shirts, y yS Figure 6.10: Production possibilities frontier poverty trap. In the figure, the country is specialized in fish and point f and producing x̄F and obtaining welfare (utility) of uF at point b, but would like to produce shirts (which would take them to a higher indifference curve). The country is better off specializing in fish than they would be if they produced a mix of goods, but they would prefer to be producing ȳS shirts at point s where they would be on the higher price line and therefore obtaining higher utility at uS at point a (uS > uF2 > uF1 ). The country cannot simply shift inputs to produce new outputs as that would require a dramatic re-purposing of production. s Price line when specializing in shirts Feasible frontier (production possibilities frontier) a c b uS Price line when specializing in fish uF2 uF1 Feasible outputs f xF Kilograms of fish, x higher levels of productivity across the board. Public policies favoring a locally dominant industry also reduce costs. If the relevant economies of scale do not pertain to an individual firm, but instead are economies of agglomeration, then a single firm even if capable of operating at a large scale will have little reason to, for example, introduce a truck manufacturing plant in a finance agglomeration such as Hong Kong, or an IT region such as the Silicon Valley. In Figure 6.10, a country could find itself at point c on indifference curve uF 1 where they have diversified production or at b where they have specialized production in fish, trade some of the fish along the price line at price p and arrive at bundle b on indifference curve uF 2 . They are better off specializing in fish than they would be if they tried to produce a diversified set of goods as shown by the specialized production resulting in higher utility at b than at c. In contrast, if the very same country specialized in producing shirts and then traded some of their output to acquire fish, they would consume at bundle a on indifference curve uS . Suppose that for as long as anyone can remember they have specialized in fishing. Why don’t they just change that? In the case of April deciding to specialize in data entry, or Brett in making graphs, the two people would have quickly realized that they were ignoring their comparative advantage: they would quickly switch their specialization to what in which they have a comparative advantage. But in the case of an entire country how would they switch? 315 316 MICROECONOMICS - DRAFT Economies of scale and poverty traps as an assurance game Player B Suppose that in the fish-producing country a few people realized that every- fish. You can see this because getting more shirts by producing them – that is moving along the feasible frontier away from point x̄F – rather than producing fish and trading them for shirts is a losing proposition. With specialization, people would produce at x̄ and trade to point b on indifference curve uF 2 But, if they clothed themselves by producing shirts rather than exchanging fish for shirts the best they could do would be point c (diversified production) on a lower indifference curve (uF 1 ). A country specializing in fish in this model is locked in to lower income. If they all decided to switch then they would all be better off. But as long as the decision about what each person will produce is taken independently, people would not specialize in shirt production. They are facing a coordination problem similar to those discussed in Chapters 1, 4, and 5. To see this, imagine that the population of the country we have been modelling is composed of just two people Anjali and Budi. Budi’s parents have urged him to take up fishing, and Anjali’s parents, too, have urged her to continue with the family’s traditional livelihood. To determine if each will take up fishing or shirt making they will engage in the non-cooperative game shown in Figure 6.11. Assuming that each spend 5 hours a day working, we have calculated their output depending on their choice and the choice of the other, using the production functions in Figure 6.4. So: • Fishing for 5 hours will produce 2.5 kg independently of what the other does, and the price of fish is 1, so the value of their output if either of them fish is just 2.5. • If both produce shirts, that is 10 total hours of shirt production resulting in 10 shirts, or 5 for each of them; at the price 0.67, they both receive a value of output of 3.33 (5 ⇥ 0.67) • If one produces shirts and the other does not, the production function for 1 s 2 shirts tells us that the output will be 10 (l ) , with l s = 5, this results in 2.5 shirts, with a value of 1.67 (2.5 ⇥ 0.67). You can use the circle and dot method (introduced in Chapter 1) to identify the Nash equilibria of the game. There are two: both fish or both produce shirts, and producing shirts Pareto dominates fishing. How would the two play the game? That would depend on their beliefs. Budi Shirts produce shirts they would be much worse off than specializing in producing Player A specializing in shirts. What could they do? If people decided individually to 3.33 Fish Shirts one would be better off (be on a higher indifference curve) if they switched to 2.5 Fish 3.33 2.5 1.67 1.67 2.5 2.5 Figure 6.11: Two people in a country have to choose whether they will play "Shirts" or "Fish". At given prices of p = $1 for fish and p = $0.67 for a shirt, they confront the following payoffs. The game is an assurance game with two Nash equilibria (Shirts, Shirts) and (Fish, Fish) where (Shirts, Shirts) is Pareto superior to (Fish, Fish) and Pareto efficient. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 317 might reason that for him taking up shirt making is risky because if Anjali does not make the same choice then there will be no economies of agglomeration and the payoff would 1.67. Fishing by contrast is a sure thing: 2.5. Anjali might well think the same way. Based on the traditions of their society they would probably believe that the other would take up fishing. And so they would both fish. Of course if they could have agreed to both produce shirts, then they would have benefited from the economies of agglomeration and each produced twice as many shirts (5) as one of them working singly could do. But we are letting Anjali and Budi represent an entire population who are mostly strangers to one another, not two neighbors who could agree on a course of action. So they have no way of coordinating their actions. Like the farmers of Palanpur – all planting late when they could all be better off by planting early – they will be less well off because of the poverty trap which they cannot escape because they lack institutions that would coordinate a joint decision. This kind of self-perpetuating specialization is part of the reason why so much of the world remained poor while other parts became wealthier. The labor force of Africa, Asia and Latin America engaged in agriculture and other low productivity sectors. Europe and its offshoots (North America, Australia, and New Zealand) became wealthier starting in the early 19th century in some measure by producing shirts and other manufactured goods. In the late 19th century and into the 20th century other countries shifted their specialization to sectors with higher labor productivity. This began with Japan, and continued with South Korea, Singapore, China, and Vietnam. These countries shifting to manufacturing as a higher labor productivity activity mirrors our example of switching to shirts from agriculture. The modern manufacturing in these countries includes electronics, ship-building, and automobile production. In all of these cases the change in specialization occurred as a deliberate government project, not as the result of countless people deciding to produce commodities like shirts rather than fish. 6.8 Application: The limits of specialization and comparative advantage Economies of scale and opportunities to exchange are pervasive in modern capitalist societies, and, as a result, we live with an extensive (even global) division of labor in which many individual households and firms specialize in producing only one or a narrow range of products and meet their needs by exchanging these products through monetary transactions. Figure 6.12: A street-side dosa. Courtesy Sachin Gupta. CC ShareAlike. 318 MICROECONOMICS - DRAFT When we think of specialization, we often conjure images of Silicon Valley’s engineering and technology hub or the City of London financial center. But, India is home to one of the most developed and specialized information technology industries in the world based in Bangalore. The Bangalore based IT firms InfoSys and Wipro exemplify the dynamics of an industry that grew from nothing in the early 1980s to become major global players by the early 2000s.2 Specialization occurs, too, in older industries. Manufacturers in Bangladesh export a lot of shirts and hats, and very few bed sheets, whereas firms in Pakistan export a great number of bedsheets, but very few hats.3 Neither is a particularly skill-intensive kind of production and there is no reason for us to expect that one of them ought to be better at bedsheets than hats. But they have specialized due to the advantages of learning by doing and economies of scale. In contrast with this specialization, however, many households do still remain diversified rather than specialized. Many households cannot achieve the benefits of economies of scale due to insufficient wealth to sustain the training and investment required for specialization, and also because of the riskiness of starting businesses or engaging in just a single kind of work. As a result, many poor households diversify of rather than specialize. For example, Abhijit Banerjee and Esther Duflo describe the economic lives of poor women in Guntur, a city in India.4 The women spend time in the morning selling dosas (a rice and bean breakfast food), they make small amounts of money collecting trash, they gather firewood to sell, they sell fruit, vegetables and clothing (mostly saris), they make and sell pickles, or they work as shortterm laborers. Similar patterns of diverse occupations occur in Cote d’Ivoire, Guatemala, Indonesia, Pakistan, Nicaragua, Panama, Timor Leste, and Mexico. An example from India shows one extreme: a survey by Nirmala Banerjee in West Bengal showed that the average family had three people who worked, sharing seven occupations among them.5 The economic analysis of these two different configurations of production – specialization or diversification – is based on the same fundamental concept – doing the best you can given a set of constraints. But as the examples above show, whether a person or family specializes or diversifies is not simply a matter of technology – economies or diseconomies of scale for example, or learning by doing or differential skills. For a family with limited or no wealth and exposed to uncertainty of their incomes in any single pursuit, risk mitigation becomes an important priority. As a result, diversification may be the best they can do. We show in Chapter 13 how this very common combination of limited wealth and exposure to uncertainty may contribute to the perpetuation of poverty. E X A M P L E Abhijit Banerjee and Esther Dulfo won the Nobel Prize in Economics in 2019 (alongside Michael Kremer). They won the prize for their work on projects to alleviate poverty using the methods of randomized controlled trials, which they have advocated for worldwide to understand the impacts of policy. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 6.9 Production technologies In modern economies, production takes place in families, in governments, in privately owned firms and in other settings, each distinguished by a character- 319 P RODUCTION Production is the process by which we transform the resources of the natural world using already produced tools, facilities, and inputs to meet human needs. istic set of rules of the game determining who owns the goods produced, who directs the production process and so on. Private owners of the buildings, machinery, intellectual property and other assets making up a firm aim to sell the output (which they also own) for more than their inputs cost, the difference between sales revenues and costs being the owners’ profit. The owners (or managers) of the firm choose the methods of production, the amounts of inputs it hires (hours of labor, number of machines), and the level of output to maximize their profits, given the production methods available, the prices they pay for inputs, and the market prices for their output. For this reason owners of firms want to minimize the costs that they incur to produce any given level of output that they decide to produce. Here we explain how they choose cost-minimizing technologies to use in converting raw materials and some given level of output of products for sale. In Chapters 8 and 9 we turn to the owners’ decision about how much to Inputs and outputs Consider a firm producing an output, for example, cars, smartphones, or clothing. To produce its output, x, the firm needs to hire labor with the skills necessary for the production tasks and provide the workers with raw materials, tools and facilities. In a general model we could think of the inputs as a list describing the amounts of labor of each kind and of all the different raw materials (wood, steel, plastic, glass), tools (dies, drill presses, forges), and facilities (factories, vehicles) required to produce the output. These inputs to the production process are sometimes termed factors of production. We would measure all these inputs over the same time period as output: so many hours of each kind of employee per month, so much steel per month, so much factory space per month, and so on. It’s easy to see how the process works if we look at two dimensions on horizontal and vertical axes. The labor hired is l (on the horizontal axis), and k (on the vertical axis) is the quantity of capital goods that the firm uses – the machines, tools and facilities that the firm needs to hire or own to produce their output over the relevant time period. We can describe one way of producing a particular level of output by indicating in this space a level of the labor input l , and the capital goods input k that will produce the specified output. This combination (x, l, k ) describes one of the possible firm’s technique of production. For a given level of x, we can describe the technique of production Quantity of capital goods, k produce. k1 A production technique, (x, l, k) i l1 Hours of labor, l Figure 6.13: A production technique, (x, l, k ) producing some amount of output x using labor l1 and capital goods k1 . FACTOR OF PRODUCTION Any input into a production process is called a factor of production. In the past economists often referred to land, labor and capital goods as primary factors of production, but this usage is outdated given the essential role today of other production inputs such as our natural environment beyond "land" and knowledge. T ECHNIQUE OF PRODUCTION A technique of production is a particular way of producing some given amount of output (x). In this case it is a combination of an output level, hours of labor input, and capital goods input, (x, l, k). T ECHNICAL EFFICIENCY A technique of production is technically efficient if there is no other technique with which the same output can be produced with less of one input and not more of any input. 320 MICROECONOMICS - DRAFT Quantity of capital goods, k Feasible k2 ● k1 ● f i ● h Figure 6.14: Production techniques. For a given level of output x = 100, we can describe any technique of production as a point showing the amount of labor, l , and the quantity of capital goods, k, required to produce output x. The area shaded in green shows the feasible combinations of labor and capital goods that can produce output x, which is equivalent to the feasible set introduced in chapter 3. The area in blue shows the infeasible combinations of capital goods and labor to obtain an output of x = 100. g ● Infeasible l1 l2 Hours of labor, l as a point in (l, k ) space, as in Figure 6.13. Technology and feasible production The firm is constrained by the available technology, which describes what techniques it can in fact carry out, given its state of knowledge, the skills of workers, and the conditions of work (health, safety and intensity) that the firm can legally and socially impose on its workers. Technology is therefore not just a question of engineering or scientific knowledge, but also involves relations between workers and management and among workers, and the legal and institutional framework within which the firm operates. Figure 6.14 displays the feasible set for producing a hundred units of output. The green shaded region shows the set combinations of capital goods and labor, sufficient to produce a given output, x = 100, of the good x. The dark green line is the border of the feasible set and is called an an isoquant. There are additional isoquants each associated with the differing level of output that the inputs produce. So there is a set of isoquants, called an isoquant map, each one of them derived from a production function and associated with a different level of output. Each of the four lettered points in Figure 6.14 is a particular combination of labor and capital goods that are sufficient to produce 100 units of good x. But the owners of a firm seeking to produce that amount would not be equally happy to use any of the four. I SOQUANT An isoquant gives the combinations of two inputs that are just sufficient to produce a given level of output. ‘Same quantity’ is exactly what the two parts of the name isoquant mean: ‘iso’ for ‘same’ and ‘quant’ for quantity. The quantity that is the same on the production isoquant is the quantity of output. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 321 k3 Steeper ray: a● more k−intensive less l−intensive k1 ● b Flatter ray: more l−intensive less k−intensive l1 xB = 100 Quantity of capital goods, k Quantity of capital goods, k xA = 100 k3 c combines techniques a and b a● k2 c ● k1 l3 ● b l1 l2 l3 Hours of labor, l Hours of labor, l (a) Comparing two techniques (b) Combining two techniques Technique i dominates technique h because it uses less of both inputs to produce the same output. Similarly techniques f and g are dominated by point i: each use the same amount as does technique i of one input and more of the other to produce 100 units. Technique i is called technically efficient because (considering the alternatives, f, g, and h) there is no other technique that produces the required amount of output (x = 100) with less of one input and not more of any other. The isoquant map derived from a production function is analogous to the indifference curves based on the utility functions in Chapter 3. It is important to remember that an production isoquant is a constraint on the choice of inputs required to produce a particular level of output, rather than something to be maximized. A utility maximizer wants to get to the highest possible indifference curve given the set of feasible options. The cost-minimizing firm wants to get to the minimum cost point on the production isoquant for any given level of output. Figure 6.15 shows a production isoquant with two techniques of production one of which uses more capital goods and less labor than the other. The production isoquant includes the points representing the two techniques and the line joining them, representing the possibility of doing some of the production with one technique and some with the other. The second technique of production provides some possibility of substitution of one input for the other by switching from one technique to the other, but this substitution is limited because there are only two techniques. x = 100 Figure 6.15: Production isoquant combining two techniques For a given level of output x, there may be more than one feasible technique of production. The production isoquant in this figure consists of the two techniques, and the straight line between them, representing production with a combination of the two techniques (as shown by point c). The availability of more than one technique implies that substitution of one input for the other is possible by shifting some production from a more capitalintensive technique to a more labor-intensive technique. R E M I N D E R A Pareto-efficient allocation is one that is not dominated by any alternative, so there is no other allocation that is preferred by at least one person and not "dis-preferred" by any person. The definition of technical efficiency is similar but applying to techniques and inputs used rather than allocation s of goods, and people’s utilities. 322 MICROECONOMICS - DRAFT As shown in Figure 6.14 and 6.15, the production isoquant can be thought of as points corresponding to the various techniques of production, and the lines connecting those points (which correspond to mixing the techniques of production). The production isoquant is equivalent to the idea of the feasible frontier in earlier chapters as it defines what combinations of labor and capital goods can feasibly produce the level of output, x. 6.10 Production functions with more than one input The techniques of production available are often described in a production function, which is a mathematical expression giving the least quantity of inputs – such as capital goods (k) and labor (l ) – that are sufficient to produce any given level of output, x. The production function can also be thought of as specifying the maximum level of output attainable for each combination of inputs: Production Function x f (l, k) = (6.10) You have already seen examples of the simplest production function in the Leontief production function in which there is but a single technique available for a given level of output x as in Figure 6.14 and Figure 6.16. R E M I N D E R We have already examined production functions, but so far they have only involved one input, such as labor as an input into studying in Chapter 3 or labor as an input into either fishing or shirt production in section 6.2. P RODUCTION FUNCTION A production function x = f (l, k ) describes a firm’s available set of techniques of production as a mathematical relationship. Here we present production functions with just two inputs – labor and capital goods – but production functions may describe the relationship between output and any number of inputs, labor with different skills, for example, or different kinds of capital goods (buildings, machines, and so on). As in those figures what are called Leontief production isoquants are rectangular because the technology specifies a given ratio of capital goods to labor (at the point of the rectangular isoquant). If that particular ratio of inputs is in use, then adding more labor or more capital goods has no effect on production: their marginal products are zero. There are therefore no possibilities of substituting one factor of production for another. To clarify what this "no substitution" assumption means with an extreme example, think about nuts and bolts: if you have n nuts and n bolts, then having n + 1 bolts is no better than having n bolts. A bolt is useless without a nut, and a nut is useless without a bolt. You need to use the inputs in fixed proportion to each other to get “a nut and a bolt.” M-Note 6.3: Leontief Production Function The output of good x, is produced with l the amount of labor input used and k the amount of capital goods used. al and ak are the minimum amounts of labor and capital goods required to produce a single unit of output. Noting that min(m, n) means m and/or n, whichever of m or n is least (or both of them if they are equal), the Leontief production function can be written: x = f (l, k) = min ✓ l k , al ak ◆ (6.11) The equation can be read: "The number of units of x produced is the smaller ("min") of the ratio of the amount of the input used (the numerator in the two fractions) to the input required for a single unit of production (the denominator) Any capital goods input in excess H I S TO RY Wassily Leontief (1906-1999) was a Russian-American Nobel Laureate in economics. He modeled the whole economy as what became known as an input-output system, with each industry being represented by a Leontief production functions. His work is valued by economists because it allowed a mathematical representation of the whole economy that could be estimated empirically (for example, engineers could determine how many tons of coal are needed to produce a ton of steel.) P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 323 of the minimum amounts required is of no use in production, and might as well be thrown away, and similarly for any labor input. Checkpoint 6.5: Leontief Production a. Using the Leontief production function in the M-Note, if al = 2 hours and ak = 1 hour of machine use, what is the level of output in each of the following cases: i. l = 10 and k = 10 E X A M P L E Leontief’s input-output models are today used, for example, to calculate the amount of CO2 emissions produced per unit of output of each industry, taking account of both the direct and the indirect inputs. That is counting for example not only the coal used to produce a ton of steel, but the coal used in producing the machinery and all of the other inputs required for a ton of steel.6 ii. l = 10 and k = 5 iii. l = 16 and k = 5 b. In each case above (i, ii, iii) how would output change if one more hour of labor or one more unit of machine time were devoted to production (this is the marginal product of labor and of machine time, respectively)? Cobb-Douglas production function Another representation of how inputs are combined to produce outputs is the Cobb-Douglas production function. Cobb-Douglas Production Function x(l, k) = ql a kb R E M I N D E R The Cobb-Douglas production function has the same structure as the Cobb-Douglas utility functions we studied in Chapter 3. (6.12) The Cobb-Douglas production function requires that l > 0, k > 0 for production to take place: some of both inputs are essential, but their proportions used can vary. • 0 < a and 0 < b capture the contribution of labor and capital goods, respectively to producing output; • The sum of a and b tells us how output responds to changes in propor- H I S TO RY Paul Douglas (1892-1976) developed the function with his colleague at Amherst College, Charles Cobb. Though a Quaker, Douglas was fiercely anti-fascist and during World War II volunteered for the U.S. Marine Corps as a private at the age of 50. He later won two purple hearts in recognition of the battle wounds he suffered in the Pacific theatre. He went on to be a prominent member of the Democratic Party and a U.S. Senator serving from 1949-1967. tional increases in both of the inputs indicating whether the firm experiences economies of scale, diseconomies of scale, or constant returns to scale. • q > 0 is a positive constant that captures a level of productivity of the specific technology . A Cobb-Douglas isoquant for x = 100 is shown in Figure 6.17. The negative of the slope of an isoquant at any point is the ratio of the marginal products of labor and capital goods inputs, and is called the marginal rate of technical substitution, or mrts(l, k ). The marginal rate of technical substitution The negative of the slope of a production isoquant shows the ratio in which the two inputs can be substituted for each other while output remains constant, the marginal rate of technical substitution between the inputs. The M ARGINAL RATE OF TECHNICAL SUBSTI TUTION The marginal rate of technical substitution is the rate at which labor and capital goods inputs can be substituted holding constant firm output. It is the negative of the slope of the production isoquant and equal to the ratio of the marginal products of the inputs. 324 MICROECONOMICS - DRAFT Figure 6.16: Production isoquant given a CobbDouglas production function. The feasible set of production for a given x is the set of techniques of production, combinations of labor input and capital goods, (l, k ) that permit the firm to produce x. Given the Cobb-Douglas production function x = ql a kb we can isolate k to find an equation 10 Quantity of capital goods, k 9 8 x feasible with (l, k) 7 for the production isoquant: k = ⇣ x ql a ⌘1 b production isoquant has a negative slope. 6 Cobb−Douglas isoquant 5 4 x = f(l, k) x infeasible 3 with (l, k) 2 1 0 0 1 2 3 4 5 6 7 8 9 10 Hours of labor, l marginal rate of technical substitution based on the isoquant is analogous to the marginal rate of substitution, the negative of the slope of an indifference curve, which we introduced in Chapter 3. But notice that in the case of the firm seeking to minimize the cost of producing a given level of output, the isoquant is the co nstraint not the firm’s objective. It tells the owners of the firm what combinations of inputs will produce the given level of output (the constraint). M-Note 6.4: The marginal rate of technical substitution and marginal products The production isoquant is defined as the combination of inputs that can produce a given output, f (l, k ) = x. To find the slope of an isoquant we proceed as we did when finding the slope of an indifference curve. We use the property of the isoquant that the points on it made up of different amounts of l and k result in the same level of output x. So for small changes in l and k the following is true: dx = fl (l, k)dl + fk (l, k)dk = 0 (6.13) Because along a production isoquant the difference in output is zero (just like along an indifference curve in earlier chapters the difference in utility is zero), Equation 6.13 can be understood as follows: fl (l, k)dl + fk (l, k)dk = 0 | {z } | {z } Change in x as l changes which we can rearrange as mrts(l, k) = Change in x as & k changes dk f (l, k) = l dl fk (l, k) (6.14) . The P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N Figure 6.17: A production isoquant for the Cobb-Douglas production function f (l, k ) = l 0.5 k0.5 for the output level x = 4. The marginal rate of technical substitution is the ratio of the marginal x mp products, mp = x l , which is the negative of the 10 Quantity of capital goods, k 9 8 mrts(l, k) = ● a k 8 =4 2 mrts(l, k) = 4 =1 4 5 4 mrts(l, k) = ● b 2 = 0.25 8 3 2 ● c Cobb−Douglas isoquant x = f(l, k) 1 0 0 1 2 3 4 5 6 7 8 9 10 Hours of labor, l Equation 6.14 can be stated as: Marginal rate of technical substitution = Marginal product of labor Marginal product of capital This is the negative of the slope of the production isoquant. Checkpoint 6.6: Isoquants and marginal rate of technical substitution Using the same production function as Figure 6.17, calculate a. the marginal product of labor and the marginal product of capital b. determine the mrts(l, k ) c. choose 3 different points (not a, b and c) along the production isoquant at which to evaluate the mrts(l, k ) to confirm that mrts(l, k ) decreases as l increases. M-Note 6.5: Cobb-Douglas economies of scale We start with the following Cobb-Douglas production function: x(l, k) = k slope of the production isoquant, dk dl . Three points along the isoquant curve are shown: a, b, and c illustrating how the marginal rate of technical substitution decreases moving rightwards down the production isoquant from l = 4, to l = 1 to l = 14 . Three gray dashed lines are tangent to the production isoquant, the slopes of which are the marginal rate of technical substitution, mrts(l, k ), at each point. 7 6 325 ql a kb To confirm what a firm’s economies of scale are we need to increase both inputs by some proportion, S. Therefore, increase l and k by the proportion S. That is, multiply each input by S before raising the input to the relevant power: 326 MICROECONOMICS - DRAFT x(Sl, Sk) = q(Sl )a (Sk)b = qSa l a Sb kb = Sa +b ql a kb = Sa +b x(l, k) Now, take each S out of the parentheses: x(Sl, Sk) The final step occurs because we know that ql a kb is equal to our original production function, x. If a + b is greater than one, the output grows more than proportionally with an increase of l and k by the proportion S. Therefore, the production function has increasing returns to scale. If a + b is lower than one, the production function has decreasing returns to scale. If a + b is equal to one, it has constant returns to scale. Diminishing marginal products of inputs It is important not to confuse economies and diseconomies of scale, which describe what happens when all inputs are changed proportionally with diminishing or increasing marginal productivity of one input when the others are held constant (say, increasing labor, holding capital goods inputs constant). A production function may have diminishing marginal productivity to any one input when the others are held constant, and still exhibit economies of scale when all the inputs are changed together. M-Note 6.6: Diminishing marginal productivity To compute the marginal product of labor in the Cobb-Douglas production function we start with the production function: x(l, k) = ql a kb To find the marginal product of labor, we calculate the first partial derivative of the production function with respect to labor, which gives us the effect on total output of a small change in the labor input, holding constant the level of capital goods input: ∂ x(l, k) = xl = MPl ∂l = = = aql a 1 b k aql a kb l ax(l, k) l For a > 0 and l > 0, the marginal product of labor is positive: as you can see from the equation immediately above, it is equal to a itself times the average. To work out whether the marginal product of labor is diminishing, we need to know whether the derivative of the marginal product of labor with respect to the labor input itself M - C H E C K For the Leontief production function we cannot compute the marginal rate of technical substitution from the slope of a production isoquant, because it’s slope is undefined at the kink in the isoquant. But at the "kink" in the isoquant, adding more capital goods or more labor has no effect on output, so we could view the Leontief production isoquant as representing an extreme form of diminishing marginal products. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N is positive, zero, or negative: ∂ 2 x(l, k) = xll ∂ l2 = a (a = = 1)ql a 2 b k 1)ql a kb l2 a (a 1)x(l, k) l2 a (a The sign of xll depends on the size of a . Diminishing If a < 1, xll < 0 because a ductivity of labor. Constant If a = 1, then a tivity of labor. 1 < 0, which implies diminishing marginal pro- 1 = 0, and xll = 0, which implies constant marginal produc- Increasing If a > 1, xll > 0 because a tivity of labor 1 > 0, which imply increasing marginal produc- Checkpoint 6.7: Marginal products of factor inputs Check your understanding by doing the following: a. Repeat the steps in M-Note 6.6 to find the marginal product of capital goods with a Cobb-Douglas function. b. With b = 0.4 is the marginal product of capital goods diminishing, constant or increasing? c. Determine the values of b under which the marginal product of capital goods will be diminishing. 6.11 Cost-minimizing technologies Having introduced a description of the production process – the production function – we now introduce the firm as a profit-maximizing entity. To determine the level of output that will yield the greatest profit for the owners of the firm, consider two pieces of information that the owners of the firm would need: • Cost minimization: for every possible level of output, given the costs of using the inputs to the production function, find the technique of production that minimizes the costs of production; • Profit maximization: using the resulting cost curve (describing the least cost at which each level of output can be produced) and the demand curve for the firm’s product, determine the level of output to produce. Here we describe cost minimization. We describe profit maximization step in Chapters 8 and 9. We call any particular combination of labor and capital goods used (l, k ) as a bundle of inputs. Finding the minimum cost bundle for producing each level of output the firm’s owners might want to produce requires three steps: 327 328 MICROECONOMICS - DRAFT Figure 6.18: Three isocost lines are presented: c1 , c2 and c3 . Isocost curves closer to the origin are made up of less costly input bundles. The equation for an isocost curve is given by c = pk k + wl , where pk is the cost per unit of renting capital goods, k is capital goods input, w is the wage, and l is the quantity of labor input. We can re-arrange this equation⇣in terms of the capital ⌘ Quantity of capital goods, k c3 pk c2 pk Marginal rate of transformation: w mrt(l, k) = pk goods input, k = c pk w pk l . The slope of the isocost line is determined by the marginal rate of transformation of capital goods into labor, mrt (l, k) = pw = dk dl , which is the opportunity cost c1 pk k of using more labor in terms of the lesser quantity of capital goods that can be used, in order to hold constant the cost of the resulting bundle. c1 c2 c1 w c3 c2 w c3 w Hours of labor, l • Step 1: Calculate the cost of every input bundle that the firm might use • Step 2: Identify bundles that cost the same, and use the resulting isocost line to distinguish between more costly and less costly bundles and • Step 3: Use the isoquants based on the available production functions to determine, for each level of output, the least costly bundle. Isocosts: Equally costly bundles of inputs We assume that: • the capital goods used by the firm are rented (for example, buildings and equipment) rather than owned; and • the firm’s own demand for labor and capital goods does not influence the price it pays for these inputs (as would be the case, if the firm is small in relation to the markets for its inputs, labor and various types of capital goods). Then the cost using any particular combination of labor and capital goods depends on: • Wages (w) paid per hour for the for the hours of labor hired (l ) • The rental cost of the capital goods ( pk ) times for the quantity of capital goods used (k). Then the cost of a bundle of inputs is: c(l, k) = wl + pk k (6.15) Quantity of capital goods, k P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N a ka 329 Figure 6.19: The minimum cost of producing a given level of output. To produce the output given by the isoquant x, the least cost input bundle is indicated by point i. mrts(l, k) = mrt(l, k) mpl w = mpk pk b kb d kd c1 la isoquant x=x c3 c2 lb ld Hours of labor, l Using Equation 8.2, we know the cost of every input bundle so we can construct an isocost line, a line showing all the possible combinations of amounts of labor and amounts of the capital good that result in a constant or equal (“iso”) level of costs. Re-arranging equation 8.2, we can find the equation for an isocost line: Isocost line k = c pk ✓ w pk ◆ l (6.16) The isocost lines represent the objectives of the owners of the firm. The owners would like to find the way of producing their product that (for some given amount of output) will put them on the lowest isocost line, that is, the one close to the origin. The constraint limiting the owners decision is the available technology or technologies as described by the production isoquant I SOCOST LINE The line through an input bundle (l, k ) when the wage is w and the price to hire capital goods is pk is the line through the input point with slope equal to w/pk , and represents all the input combinations that have the same cost. R E M I N D E R In earlier chapters the indifference curves bowed in towards the origin (like the green isoquant in Figure 6.19) represented the objectives of the person, that is the thing that she wished to maximize based on their preferences, subject to some constraint, for example a limit on how much she could spend. Here the blue isocost lines represent the objective of the firm’s owners, that is the thing they wish to minimize, while the curved isoquant is the constraint based on the feasible set of production techniques that produce at least the given amount of output. for the given level of output. The general principle of cost minimization Contrast point b with points a and d. At a, the marginal rate of technical substitution is high (the isoquant is steep), meaning that it can reduce capital goods inputs substantially and still sustain the same level of output with only a modest addition of the labor input. At the going cost of renting capital goods and the wage rate and the firm would decrease its costs if it employed fewer capital goods and more labor. The effect of this is to lower the marginal product of labor and raise the marginal product of capital goods. It would continue to substitute labor for capital goods until the point where the ratio of marginal products equals the ratio of prices for the inputs at b. The isocost line and production isoquant toolset allows us to understand cost P RINCIPLE OF C OST M INIMIZATION A firm with a production isoquant consisting of a continuum of techniques of production defined by a production function x = f (l, k ) will minimize its costs at the point where its marginal rate of technical substitution of f capital goods for labor, (mrts(l, k ) = f l ), k equals its marginal rate of transformation, or the price ratio of labor for capital goods, mrt (l, k) = pw . The principle of cost mink imization is satisfied where the production isoquant is tangent to the lowest isocost line. 330 MICROECONOMICS - DRAFT minimization. The firm wants to choose the lowest possible isocost line to produce the output necessary to produce the target output, the minimum cost technique of production. Minimizing the cost of producing some hypothetical level of output involves the principle of constrained optimization from Chapter 3. The constraint is the isoquant for this particular level of output and the objective is to reach the lowest isocost line. Parallel to the principle of demand in chapter 3 for people maximizing utility, we have a principle of cost-minimization for firms choosing techniques of production. The firm will produce where mrts(l, k ) = mrt (l, k ) or where the f ratio of marginal products equals the ratio of input prices, f l = pw . k k The cost-minimization graph and the graph describing the utility-maximizing choice of a consumer facing a budget constraint are similar. But it is important to recognize that the meaning of the elements is different. The constraint in the production case is that the firm produce some given amount x = x, not the isocost line. The owners of the firm are trying to move as close to the origin as possible within the constraint that it produces the specified amount, while the consumer is trying to move to as high an indifference curve as possible within the constraint of the budget available. The iso-cost lines are analogous to the indifference curves of the consumer (costs are being minimized, like utility was being maximized), and the production isoquant is analogous to the feasible frontier (it is the constraint). Checkpoint 6.8: Choices of capital goods and hours of labor Make sure you understand Figure 6.19 by explaining why, if the firm were producing at point d, it could reduce costs of producing the given amount of output by using more capital goods and less labor. Input prices and the choice of a labor-intensive or a capital-intensive technology Now, think about a firm that sells some product and is considering which of two technologies to use producing it. One uses some powerful machinery (the capital good) and little labor while the other technology uses lots of labor and a smaller machine. For concreteness, think of the two technologies as similar to plowing a field using a powerful tractor or with a small garden type roto-tiller. If these two alternative ways of producing the good were described by a Leontief technology, then we could say that the one using the roto-tiller is the more labor intensive, or what is the same thing (because there are just two inputs) the less capital goods intensive. In the Leontief technology, the ratio P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N Isoquant F xf(lf, kf) = x ki Isocost curves (high wage) cH 1 kf Isoquant G xg(lf, kf) = x Quantity of capital goods, k Quantity of capital goods, k cH 2 f Isocost curves (low wage) kg i mrtsB(l, k) > mrtsA(l, k) xBl cHʹ 1 xBk kh j kj xAk lh of output, that is the ratio of the amounts al /ak , is a measure of the labor intensity of the technology. While the more accurate expression is to refer to capital goods intensive technologies, to save words we sometimes refer to technologies as "capital-intensive." Figure 6.20 a. illustrates this case first with Leontief technologies and Figure 6.20 b illustrates Cobb-Douglas technologies. In Figure 6.20 b, the CobbDouglas technology indicated by point g is more labor-intensive than the technology at point f. Where substitution between inputs is possible – as with the Cobb-Douglas technology the distinction between labor-intensive and capital-intensive techintensive technology is the one that the owners of a firm would choose to would minimize costs if wages were low relative to the cost of capital goods. A capital-intensive technology, analogously, is one that would be used by a cost-minimizing firm if wages were high relative to the costs of capital goods. Point f in Figure 6.20 a. shows the inputs required to produce a single unit of output using the capital-intensive technology. Point g shows the same information for the labor-intensive technology. Which technology the firm will adopt in order to produce its product at the lowest cost depends on the relative cost of labor and capital goods, as indicated by the isocost lines in li lj Hours of labor, l (b) Two Cobb-Douglas technologies of inputs of labor to the inputs of the capital good required to produce a unit nology is not so simple. The basic idea, however, is the same: the labor- Labor−intensive technology B xB⎛⎝l, k⎞⎠ = x cLʹ 1 cL2 (a) Two Leontief technologies green and blue. xAl Capital−intensive technology A xA⎛⎝l, k⎞⎠ = x lg Hours of labor, l lf > h g cL1 331 Figure 6.20: Choosing a capital-intensive or labor-intensive technology to minimize costs. The cost-minimizing choice of technology depends on the wage and the cost of capital goods. Higher wages (a steeper blue isocost lines) will lead the owner to implement the more capital goods intensive technology. In panel b the coefficients for the labor-intensive Cobb-Douglas technology B are a = 2/3, b = 1/3, and for the more capital goods-intensive Cobb-Douglas technology, A a = 1/3, b = 2/3. 332 MICROECONOMICS - DRAFT If wages are low, then the isocost lines are flatter, as shown in the figure with the green isocost lines. If the firm uses the labor-intensive technology it will incur costs of cL1 which is less than the cost it would incur if it used the capitalintensive technology when there are low wages (along cL2 ). Higher wages (for the same rental cost of the capital good) are indicated by the steeper isocost lines in blue. Using the labor-intensive technology with higher wages (along cK 2 ) would incur higher costs than using the capitalintensive technology (along cK 1 ). Figure 6.20 b. shows an analogous situation with greater substitutability between the two factors of production with Cobb-Douglas technologies. Once K0 again, the relative costs are shown by two iso-cost lines, cL0 1 and c1 . The unit isoquants show the different combinations of capital goods and labor that would produce the same output, x. The owners of the firm would choose point h if wages were high and point j if wages were low. Along the ray going through points h and i, the ratio of kl shows that Technology A is capital- intensive. At points h and i, the technical rate of substitution differs between the two isoquants (which we can see with the labor-intensive technology B having a much steeper isoquant at point i than the capital-intensive technology has at point h). Checkpoint 6.9: Capital-intensive and labor-intensive technologies • In Figure 6.20 panel a, show that there is one ratio of wages to the cost of capital goods such that the least cost of producing x will be the same using the two technologies. • Show that the input price ratio the of firm using technology F will use less labor and more capital goods than the firm using technology G. • Show that if a firm had just two technologies to choose from, the Leontief technology F from panel a and the Cobb-Douglas technology A from panel b, it would choose the Cobb-Douglas technology if wages were either very high relative to the cost of capital or very low. But for some input price ratio in between, it would choose the Leontief technology. • Explain why this means that it is not always possible to designate a technology as more labor-intensive or more capital-intensive. 6.12 Technical change and innovation rents Firms and their owners compete not just by adjusting output levels but also by seeking to innovate, either by finding new lower-cost techniques of production, or by creating new products that open up new industries and new sources of demand. Quantity of capital goods, k P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N kb Figure 6.21: Isocosts and technological progress. A firm innovates to reduce its costs. The firm starts at a with its initial labor and capital goods combination (la , ka ) on isocost c2 . With innovation the firm’s constraint is eased resulting in a new production isoquant with an expanded feasible set of production. As a result, the firm can, at going prices of labor and capital goods, w and pk , employ a lower quantity of capital goods and less labor to produce output x at a lower cost total cost, moving to a lower isocost line. Following the principle of cost minimization, the firm chooses the point at which its new production isoquant is tangent to the lowest possible isocost at b, employing (lb , kb ). a ka initial isoquant x=x b c1 lb 333 innovation isoquant x=x c2 la Hours of labor, l Viable innovations The theory of cost minimization suggests an important insight into the causes of innovation in production. A new technique of production will be of interest to owners only if it lowers costs of production given current input prices, the wage, w and the rental price of capital goods, pk . The introduction of a new machine or a new organization of the production process may improve on some existing available technique of production but it will be irrelevant to the firm unless it results in lower costs than the existing minimum cost technique of production. How would we represent technological progress with production isoquant curves? With technological innovation the firm should be able to produce the same amount of output at lower total costs. We present production isoquants with technological innovation in Figure 6.21. The initial technology is shown with the cost-minimizing point a where the firm employs the combination of labor and capital goods (la , ka ). With technological progress, two things occur: • The firm is able to produce the same amount of output with fewer inputs. Its feasible set enlarges, resulting in a new production isoquant closer to the origin. • As a result of the new production isoquant, the firm will minimize its costs at existing input prices and move to a lower isocost c1 , finding the point of tangency of the isocost and the new production isoquant at point b. E X A M P L E Forbes magazine produces a list of the most innovative firms in the world (https://www.forbes.com/innovativecompanies/list/. In 2018 Netflix, Tesla (electric vehicles), Facebook, and Amazon were in the top ten as was Hindustan Unilever (a consumer goods producer and marketer in India), and Naver (selling computer and web services based in South Korea). 334 MICROECONOMICS - DRAFT Notice that the new technology has made both labor and capital goods more productive. Innovation rents Other things equal – importantly the prices of the inputs and its output – if the firm produces the same quantity of x with lesser amounts of inputs per unit of x, it would necessarily increase its profit. The firm would therefore obtain an innovation rent. This is a rent because the firm’s next best alternative – its fallback position – would be to not innovate. The innovating firm can lower its prices and capture a larger share of the market. Other firms will either make losses and exit the industry or innovate as well. As other firms imitate the innovating firm, they would experience lower costs and would compete with the innovator resulting in a lower price for the good and therefore lower economic profits for the firm that initially innovated. Innovation may result in new technologies that are more capital-intensive or more labor-intensive. If the production technology is Cobb-Douglas, then the innovation could be represented not only by an increase in q but also by new levels of a and b , which would indicate the change in the importance of the corresponding input. A firm that successfully innovates by lowering its cost of production raises its maximum profit. The firm’s increase in profit is an economic rent, or an innovation rent. It is an innovation rent because the firm’s fallback is the profit it would have made at the initial technology, such as that resulting from point a in Figure 6.21. The innovating firm will continue to obtain higher profits until its competitors adopt the same or equivalent cost-reducing technical improvements. Once firms producing identical or similar products have matched the innovator’s lower costs, if competition among firms is sufficient, some of them will reduce prices to gain a larger market share, forcing other firms to do the same. This will reduce innovation rents. 6.13 Application: What does the model of innovation miss? Our model of innovation captures essential parts of the process by which technical change revolutionizes an economy. But as this example shows, it misses important aspects too. A cluster of small firms in Sialkot, Pakistan produce about forty percent of the worlds soccer balls - 30 million of them per year - including the match balls for the 2014 World Cup. The industry is highly competitive not only among the hundred or so firms in Sialkot, but also on a world scale, with Chinese firms recently challenging the Pakistani dominance in the field. Firm owners are constantly on the lookout for ways to slash costs. As the artificial leather that the balls are made from constitute almost half the cost of a soccer ball, E X A M P L E Innovation rents play the key role in determining the profitability and survival of firms. Apple, for example, keeps ahead of its competitors by being the first to introduce important innovations like the iPad or the iPhone X (with facial recognition). Business history also provides dramatic examples, such as IBM in the 1980s where a firm that has managed to maintain innovation rents for many product cycles loses its position by misjudging the next turn of the technological revolution.7 P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 335 they are particularly on the lookout for waste-saving methods of cutting the pentagons and hexagons that make up the balls. An Italian architect and her husband, an American economist, discovered a way to cut the pentagons and hexagons from the large sheets of leather that would allow a considerable saving of leather. (Unwittingly they had "discovered" what is called a "packing" principle already known by mathematicians.) They found a tool and die-maker in Sialkot to make some test dies (a cutting tool) using the new technique, expecting that they would quickly be taken up by the cost-conscious firms. In May 2012 they gave 35 firms the new technology. They calculated that the new technology would increase profits of the companies adopting it by 10 Figure 6.22: [ 0cm]A soccer ball. Using the new technology the white hexagons and black pentagons making up the ball could be cut with less leather wasted. percent. Fifteen months later only 5 of the firms had made any substantial use of the new cutting dies.8 While the new design was easily copied and would have increased profits considerably if introduced, only one of the firms not given the new technology had copied it. The reason, it seems, is that the employees who would have used the new dies (cutters and printers) were paid piece rates, that is, the employees were paid per panel they cut. The payment method mattered because the new technology did not speed up the process of cutting, which would have increased the pay of the cutters. Instead the cost reduction came from saving leather, which would enhance the profits of the owners, but would not have benefited the workers. Because the cutters and printers did not stand to gain by saving leather, they had no interest in adopting the new technology. This was especially the case given the initial learning period in which the number of panels cut would actually be lower than before, meaning the workers would, for a short period, make less money. So they complained to their employers that the new dies did not work very well. Owners, lacking any independent way of verifying the competing claims of the Italo-American couple and their own cutters showed little interest in the new technology. Except one. One of the larger firms had a different pay system – the cutters were paid a fixed monthly salary rather than per panel that they cut. This firm purchased (and used) 32 of the new dies, apparently without resistance from the cutters. As long as none of the other firms adopted the new technology, this firm would then have been making substantial innovation rents due to the reduced cost of materials. If the competitive process worked in Sialkot that way economists think that it should, then this firm should have expanded its share of soccer ball production, eventually forcing other firms to either adopt the new technology, or to drop out. We do not know if that is what happened. Figure 6.23: A worker at the firm that adopted the new technology. 336 MICROECONOMICS - DRAFT 300 30000 200 Hours of work per 10 million lumen hours (ratio scale) Labor hours per 100 bushels of wheat (ratio scale) 10000 100 50 30 20 10 3000 1000 5 300 100 30 10 3 3 1 1840 1860 1880 1900 1920 1940 1960 1980 1800 1845 1855 1875 1895 1920 1940 1990 1992 Year (a) Improvements in agricultural technology Year (b) Improvements in light technology This case makes it clear that firms are made up of people, and the sometimes incomplete information and conflicting interests among them constitute impediments to improvements that in principle at least would allow for mutual gains to be shared among employees and owners. 6.14 Characterizing technologies and technical change Production technologies shape how we live, and ongoing changes in technologies are revolutionizing the world. The Industrial Revolution and changes in technology since have transformed the economies of Europe and North America from largely agricultural production to manufacturing and later servicebased livelihoods. Included were the shift of most work out of the home and into the factory or office, the enormous increase in the scale of production of typical firms, the widespread replacement of human labor by machines and vast increase in the quantity of goods and services available along with a decline in the amount of time in one’s lifetime spent working. Figure 6.24 shows the scale of these productive improvements for two technologies: agricultural output and light. Panel a shows the change in the the number of hours required to produce 100 bushels of wheat. In 1830, farmers needed 275 hours to produce 100 bushels of wheat or over thirty-four 8-hour workdays in contrast with merely 3 hours required to get the same amount of wheat in 1987 (and even less time today). In panel a, we show how the number of hours of labor time have decreased to obtain 100 bushels of wheat. A bushel of wheat is approximately 60 pounds or Figure 6.24: Improvements in farming and lighting technology over time. In both panels, improvements in technology show the reduced number of hours of labor required to obtain the indicated output. The vertical axis measures what is called a ratio scale so that, for example, the distance between 20 and 100 is the same as the distance between 10 and 50 (the ratio of the first to the second number is the same in both cases). This is equivalent to a logarithmic scale, so the rate of change of the measure is the slope of the lines shown. Sources: Nordhaus (1996) and Spielmaker (2018). xL3 k2 xL2 k1 xL1 xCD 3 xCD 2 xCD 1 l1 l2 337 Quantity of capital goods, k k3 Quantity of capital goods, k Quantity of capital goods, k P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N xS1 xS2 xS3 l3 Hours of labor, l (a) A Leontief Technology Hours of labor, l (b) A Cobb-Douglas Technology about 27.5 kilograms of wheat. The graph shows the amount of time required to obtain 100 bushels, decreasing from over 275 hours in 1830 to merely 3 hours in 1987 – a 90-fold improvement in productivity. In panel b, we show the amount of labor to obtain 1000 lumen hours. A lumen is a standard measure of light intensity equivalent to the light of one candle. Hours of labor, l (c) A perfect substitute technology Figure 6.25: Substitutability between labor and capital goods with different unit isoquants. Panel a. Leontief isoquants illustrating an elasticity R Esubstitution M I N D E R A production is one of of zero Paneltechnique b. Cobb-Douglas isoqaunts illustrating an elasticity substitution particular way of producing anofoutput, x, l, kof. one. Panel c. three isoquants of a technology in A production function (for example, the which the elasticity of substitution is infinite. Cobb-Douglas) describes a technology, that is, a set of techniques. The increasingly steep slope of the line in the panel b indicates an acceleration of the rate of decline in the amount of labor required to produce a given mount of light. Panel b shows a vast increase in productivity of human labor demonstrated by the decline in how much labor time was required to produce 10,000,000 lumen-hours of light. Two hundred years ago, to make even one candle required immense amounts of work to benefit from the output: a modest amount of artificial light. In the contemporary world, though, light arrives at the flick of a switch and the labor required to produce the electricity and the advanced technology of super efficient light-emitting diodes (LEDs) is measured in minutes not days or weeks. Both of these outputs – wheat and light – show the ways in which productivity has increased over time as a consequence of human ingenuity. You can see from the figure that the pace of productivity improvements is accelerating in lighting (the decline in labor required is steepening over time) and slowing down in farming. Interpreting technological change with production isoquants To understand how technologies impact how we work and live, and why new technologies continue to revolutionize our economy and society, think about five dimensions of a technology. • Economies of scale: Does increasing all inputs by a factor of S increase output by more than a factor of S? M - C H E C K A production function A is more labor-intensive than production function B if for any given ratio of wages to the price of capital goods, the cost minimizing choice of inputs will be to hire more labor hours when using A than when using B. 338 MICROECONOMICS - DRAFT • Overall productivity: For a given set of inputs how much output is produced? • Input intensity: Does the production process rely more on the input of labor (as in caring for children or the elderly, or doing scientific research), or capital goods (as in manufacturing) or natural resources, information, or some other input? Labor intensive and capital-goods intensive production is illustrated in Figure 6.20. • Complements and substitutes: If the amount of one input used increases and this raises the marginal product of the other input then the inputs are complement. Example: computer driven welding robots increase the marginal productivity of the engineers who program them. If the increase in one input used reduces the marginal product of another input, the two inputs are substitutes. Example: computer driven welding robots reduce the marginal product of manual welders (possibly to zero). • Input substitutability: Must inputs be used in some fixed proportions or can one input be substituted for another. An example of fixed proportions: a truck needs a driver, adding a second driver or a second truck does M - C H E C K Figures 6.20 a. and b. illustrated the isoquants of Cobb-Douglas and Leontief technologies, one of which is more laborintensive (less capital-intensive) than the other. You can check the input-intensity by looking at the slope of the isoquants (that is the, the marginal rate of technical substitution or the ratio of the two marginal products) for a given ratio of the inputs (that is, along a ray from the origin). Technology A (shown by x1A ) is more capital goods intensive than Technology B (shown by x1B ) because, for a given ratio of inputs, the marginal productivity of capital goods (relative to the marginal product of labor) is higher. not add much to the transportation services delivered. An example of substitution: calculations done "by hand" with pencil and paper can also be done by a computer, using more capital goods and less labor. The extent to which one input can be substituted for another in production is termed the elasticity of substitution defined as the percentage change in the minimum cost input proportions associated with a percent change in the ratio of the wage rate to the price of capital goods. The elasticity of substitution ranges from: • zero in the Leontief production function (zero change in input proportions, no matter how much the input prices change (illustrated in panel a of Figure 6.25) to • one in the Cobb-Douglas production function (if the ratio of the wage to the price of the capital good doubles, the ratio of amount of the capital good to the amount of labor used by a cost minimizing producer will double, illustrated in panel b of Figure 6.25), to • infinity where the inputs are called perfect substitutes, and using one of the two inputs, but not both will generally be cost minimizing (illustrated in panel c of Figure 6.25). The elasticity of substitution tells us how linear the isoquants are, ranging from perfectly linear (perfect substitutes), to extremely kinked (the Leontief case). Examples of how these dimensions of technologies altered ways of life in the ELASTICITY OF SUBSTITUTION is the percentage change in the minimum cost input proportions associated with a percent change in the ratio of the wage rate to the price of capital goods. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N past, and continue to do so today are given in Table 6.3. Key questions about technology today include: • Are capital goods in the form of robots substitutes for workers doing routine tasks and at the same time complements of engineers who operate them? If so, the marginal products of the two kinds of workers will diverge, possibly generating greater inequality between engineers and routine task workers. • As the ratio of capital goods to labor input continues to increase, this may depress the marginal product of capital goods and raise the marginal product of labor as would be expected if capital goods and labor are complements. This could be the basis of greater income equality between owners of capital goods and the workers they employ. But by how much? • Sectors of the economy with labor intensive production functions – such as education, security services, entertainment, child and elder care, and health services – are increasing their share of the economy at the expense of capital goods intensive sectors such as manufacturing. Will this result in greater scarcity of labor relative to capital goods and an increase in workers’ bargaining power? • Will devices and algorithms associated with artificial intelligence become a substitute or the work of engineers and other professionals, driving down their marginal products? 6.15 Conclusion Economics (as you read at the beginning) is the study of how people interact with each other and with our natural surroundings in producing and acquiring our livelihoods. The technologies studied in this chapter – summarized mathematically by production functions– describe how we can produce our livelihood by transforming nature – crops mineral resources and energy – in order to provide the goods and services that make up our standard of living. The available technologies and the ways owners of firms seek to maximize their profits by choosing techniques of production that minimize their costs of production have important effects on how we interact with each other in this process (the other part of the definition of economics). If technologies are highly efficient, then people will have the opportunity for high living standards for all. But if there are important economies of scale in production then it is likely that the economy will be dominated by a limited number of large firms whose owners may be able very disproportionate share of the high levels of production. If labor is highly productive relative to capital goods, then those who own the capital goods will not be able to earn such high incomes as those – virtually everyone – who provide labor to the system of produc- 339 340 MICROECONOMICS Characteristic - DRAFT Example: Example from Industrial Revolu- Cobb-Douglas tion Examples from today and future Economies of x = ql a kb a + b > Industrial revolution increased New technologies (e.g. 3-D print- scale economies of economies of scale leading to ers) may reduce economies of scale a + b < 1 larger firms scale but large first copy costs 1 diseconomies of (“prototyping”) imply economies scale of scale (e.g. R&D for producing a drug) Overall productiv- q ity Increases in productivity allowed for Is a long term slowdown in produc- improved living standards including tivity growth in our future? less work Labor intensity a a fell and the ratio of capital goods Labor with engineering and net- to labor input rose working skills may be replacing both capital goods & other labor Substitutes or Inputs are com- Capital goods were substitutes for Artificial intelligence (AI) may be- complements plements some kinds of labor (manual, rou- come a substitute for even highly tine) and complements for others trained engineering and other labor (engineering, design) Elasticity of sub- Elasticity of sub- For many early machining and pro- If the elasticity of substitution is low, stitution stitution = 1 duction line processes, substitution then the continuing increase in the was very limited. quantity of capital goods per worker could allow wages to rise relative to profits. tion. But technologies and the division of labor that results when people specialize according to comparative advantage is just one part of economic knowledge. Equally important are the wants and needs of people and how these are expressed in our willingness to pay for goods when they are supplied in markets. We turn to market demands next. Making connections Economies of scale and learning by doing are among the main reasons for the the division of labor and specialization, which makes important contributions to human well being; but we will see in Chapter 8 that economies of scale and learning by doing a may also limit the degree of competition in markets. Markets as a means of coordination: The opportunity to exchange of goods expands the set of feasible outcomes available to people and nations by providing facilitating specialization and the division of labor. Table 6.3: The ways in which technologies differ and why it matters. P R O D U C T I O N : T E C H N O L O G Y A N D S P E C I A L I Z AT I O N 341 External effects, coordination failures and poverty traps: The positive external effects associated with economies of agglomeration result in many possible Nash equilibrium patterns of specialization; countries may specialize in goods that keep them poorer than had they specialized in some other way. Constrained optimization: the choice of technology Minimizing cost subject to a production function constraint and maximizing utility subject to a budget constraint have many features in common. They are both examples of maximization (or minimization) under constraints. Innovation rents A firm that succeeds in finding a new technology that lowers costs of production at existing input prices can make substantial economic profits, called innovation rents, until others adopt the same or similar innovation. Important ideas specialization production possibilities frontier diversification technique of production production function production isoquant division of labor average product marginal product economies of agglomeration marginal rate of transformation relative price cost minimization equalization of marginal products and input prices marginal rate of technical substitution economies of scale constant returns to scale diseconomies of scale wages isocost line rental price of capital goods diminishing marginal productivity short run/long run technical efficiency opportunity cost of capital 342 - DRAFT MICROECONOMICS Mathematical Notation Notation Definition x, y x̄, ȳ p l k f () al ak a b q x S w pk c goods produced using labor and capital (or just labor) maximum feasible production of goods x and y, given the current technology price firm input: labor firm input: capital production function required amount of labor inputs in Leontief production function required amount of capital goods in Leontief production function intensity of labor in Cobb-Douglas production intensity of capital in Cobb-Douglas production parameter of productivity, Leontief and Cobb-Douglas production isoquant, a fixed amount of good x that can be produced by different combinations of labor and capital scale factor for increases in all inputs wage cost of renting capital goods cost function Note on superscripts: L: related to labor; K: related to capital. Discussion Questions See supplementary materials. Problems See supplementary materials. Works Cited See reference list. 7 Demand: Willingness to pay and prices DOING ECONOMICS This division of labour... is the necessary... consequence of...the propensity to truck, barter, and exchange one thing for another. It is common to all men, and to be found in no other race of animals... Nobody ever saw a dog make a fair and deliberate exchange of one bone for another with another dog. Adam Smith, The Wealth of Nations, Book 1 ch 2 Ancona is a town on the Adriatic coast of Italy southeast of Venice. It hosts one of the many daily fish markets that sell to European restaurants and fish dealers. Because fish (notoriously) spoil rapidly even with refrigeration, the price of fish on any one day depends largely on the amount of fish brought to the market that day (since none can be carried over from previous days). Economists view fish markets as a kind of ideal experiment for studying how supply and demand determine the prices at which goods are bought and sold . Figure 7.2 shows that the average daily prices and the average daily quantities of fish sold in the Ancona market: • if the price per kilogram of fish is high, the quantity of fish bought and sold is less, and • if the price per kilogram of fish is low, more kilos of fish are transacted. One explanation for the downward-sloping line in the figure summarizing this relationship is that typical buyers in the Ancona fish market will buy more This chapter will enable you to: • Apply constrained utility optimization to the the problem of demand, relating a person’s willingness to pay to their purchases of goods. • Derive a person’s demand curve from a utility function describing the person’s preferences. • Understand that consumption is often a social activity, so our preferences for particular goods (for example smoking tobacco products) often depend on what others are consuming. • Explain how people change their purchases when prices or income change. • Understand how these responses reflect both income and substitution effects and use these concepts to explain the effects of a proposed carbon tax and citizen dividend. • Use the concept of consumer surplus and understand the conditions under which it makes sense to sum the consumer surplus of many people. • Explain how market demand curves can be be derived from individual demand curves. • Use the price elasticity of demand to explain the effects of price increases for example resulting from policies such as placing a tax on sugary drinks. fish if the price is lower. Another is that the greater quantity of fish brought to market on any given day, the lower will be the average price per kilogram of fish. To understand how the price and quantity of fish purchased is determined, an essential concept is demand, as measured by the amount a person is willing purchase at any given price. For fish and other goods, knowing how Figure 7.1: Mosquito nets save lives; how widely used they are depends on the price. See Figure 7.3. Daily average price per kilogram (Euros) 344 MICROECONOMICS 12 - DRAFT ● 10 ● ● ● ● ● ● ● 8 ● ● ● ● ● ● ●● ● 6 ● 0 Figure 7.2: Prices and quantities of fish bought and sold in the Ancona market. The plot of daily average prices of fish in the Ancona market against the same day’s quantity of fish sold can be summarized by a downward-sloping curve. ● 100 ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ●●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ●● ● ● ● ● ● 200 300 400 ● ● ● ● ● ●●● ● ● ● ● ● 500 Daily quantity (kilograms) the amount purchased depends on the price is also an important piece of information in the design of economic policies. Here is an example. There are now many low-cost life-saving preventative health products such as insecticide-treated mosquito nets, tablets to eradicate parasitic stomach worms, and water purification products. In many countries in Africa, Asia and Latin America, these products prevent illness and death of their users, and also limit the spread of infectious diseases to others, but they are used sparingly if at all. Some policy-makers think these products should be provided free of charge to low-income families to encourage the use of the products. Other policy-makers disagree, suggesting that there should be a cost to acquiring these products to discourage wasteful use through better targeting of who gets the products. The question then arises: how will the take-up of the products depend on the price? Will charging even a small price significantly discourage use? Economists have conducted experiments in 8 countries to find answers to these questions. In the experiments, potential users are randomly selected to be offered the goods free or at one or more different prices. The average use of the products at each price (including zero) is then recorded. Some of their results are shown in Figure 7.3. Figure 7.3 shows that the effect of charging higher (even if very low) prices F AC T C H E C K Adam Smith’s claim in the head quote that humans are unique among all animals in our division of labor and exchange of goods is probably right about dogs. But Smith is certainly wrong about the many other species such as ants and other social insects that practice a very advanced division of labor and specialization. Different species of fish exchange services in what are termed "biological markets." D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S $6 Bednets (Kenya 2007, pregnant women) Bednets (Kenya 2007) Cement latrine slabs (Tanzania 2015) Clorin (Zambia 2006) Deworming (Kenya 2001) Plastic latrine slab (Tanzania 2015) Soap (Guatemala 2009) Water filters (Ghana 2009) Price in USD $5 $4 345 Figure 7.3: The demand for preventative health products: Take-up rates at various prices and when available for free. Our measure of demand is the take up rate, that is the fraction of the population that acquires the product (whether free or for a price). For most products the demand is substantially less when even a small price is charged compared with when the good is available for free; and higher prices are also associated with substantially less demand than lower prices. Source: Dupas and Miguel (2017). $3 $2 $1 0. 0% 10 .0 % 90 .0 % 80 .0 % 70 .0 % 60 .0 % 50 .0 % 40 .0 % 30 .0 % 20 10 .0 % $0 Household Take−up Rate was to reduce the amount of the product used, in some cases by a substantial amount. • In Zambia, for example, increasing the price of a disinfectant tablet from nine cents to twenty-five cents reduced the fraction of the population using the product from 76 to 43 percent. • Only 43 percent of a group of pregnant Kenyan women purchased insecticide treated mosquito nets when the price was 60 cents; virtually all used the nets when they were provided without charge. • A program in Kenya that had initially given away de-worming tablets for children, but later introduced a charge of thirty cents per child found that usage of the tablets fell from 75 per cent of the affected population to just 18 percent. On the basis of this information, the Poverty Action Lab at MIT, led by economics Nobel Laureates Abhijit Banerjee and Ester Duflo, suggested that there are good reasons to make these products available without charge or highly subsidized to ensure very low prices. We begin our analysis of how people spend their money – whether on fish or mosquito nets – with a basic fact: there are limits to how much a family or person can spend. R E M I N D E R Economists have researched preferences based on observing people’s behavior in real situations and in experiments. We reviewed some of the key findings from this work in Chapter 2 and we will review more experimental data in later chapters. 346 MICROECONOMICS - DRAFT Figure 7.4: Budget constraint for coffee and data. The budget set is shaded in green and the budget constraint is the dark green line on the border of the budget set. Consumption bundles (x, y) in the budget set and on the budget constraint can feasibly be obtained with the current budget (m) at going market prices for x and y, px and py . Outside the budget constraint, in the shaded blue area, the bundles of x and y cannot feasibly be obtained at going market prices with the existing budget. m y= py Gigabytes of data, y Infeasible (outside the budget) Feasible (within the budget) Budget Constraint m ⎛px⎞ y = − ⎜ ⎟x py ⎝py⎠ Kilograms of coffee, x x= m px 7.1 The budget set, indifference curves and the rules of the game. To understand how prices influence the take up of one of the life saving health products in Figure 7.3 or the amount of some good that we will consume, think about someone who has a total amount of money to spend m that she has in cash, savings, available credit, and so on. The budget constraint We shall consider a person, Harriet, and the decisions she needs to make. A person’s budget set states what bundles (x, y) are feasible for her to consume given her budget and market price of the goods: m px x + py y m Prices of Goods ⇥ Quantities of Goods Purchased (7.1) Expressing this inequality as an equality – assuming that Harriet would not consume less than her budget allowed – we have the budget constraint m Budget = px x + py y = Prices of goods ⇥ Quantities of goods purchased (7.2) (7.3) Equation 7.3 is a statement about prices and budgets. But it is also a statement about the rules of the game and preferences. They are as follows: R E M I N D E R We saw budget constraints in Chapter 3: the slope was a price ratio that represents the opportunity costs of spending money on one good in terms of how much less of the other you can afford depending on your budget. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 347 • No gifts, thefts or consumption as a matter of citizen rights: You can consume only what you pay for; so no gifts or goods provided by government, or acquired by theft. • No altruism or concerns about environmental sustainability: You consume the most that your budget allows and you consume it yourself, rather than giving it to or sharing it with others. We can rearrange the budget constraint, Equation 7.8, to obtain a line we can draw on the x and y-axes we use for indifference curves: y= m py px x py (7.4) B UDGET S ET & B UDGET C ONSTRAINT The budget set is the set of all feasible purchases of the bundles of x and y with current budget m, such that m px x + py y. The budget constraint is the border of the budget set showing all combinations that exhaust the budget, i.e. for which the constraint holds with equality, m = px x + py y. We plot the budget constraint in Figure 7.4. Examining the two terms on the right-hand side of Equation 7.4, we can see that if Harriet were to consume only good y and no good x, then she would consume pm units of good y which y is the intercept of the budget constraint with the y-axis. As Harriet buys more of good x, she moves along the budget constraint with the slope px py indicat- ing the rate at which she can sacrifice good y for good x. If she were to buy only good x, she could afford x = pm units of good x. x The negative of the slope of the budget constraint is a relative price and measures the opportunity cost of obtaining good x in terms of the amount of good y that Harriet must sacrifice because her funds are limited. The (negative of the) slope of the budget constraint is another marginal rate of transformation, it tells us the terms on which a reduced amount of good y can R ELATIVE P RICE A relative price shows the price of one or more goods relative to another good, as such it is indicated by a ratio of the one price relative to the other. Relative prices show the opportunity cost of having more of one good in terms of the lesser quantity of the other imposed by the budget constraint. be "transformed into" additional amounts of good x while just satisfying the budget constraint. M-Note 7.1: Budget for coffee and data For particular values of m, px and py we can graph the budget constraint. Consider the following example: • Harriet has a budget (m = $50) to spend on kilograms of coffee, x, and gigabytes of data, y. • The price of a kilogram of coffee, px , is $10. • The price of a gigabyte of data, py , is $5 Putting these pieces of data together, therefore, the budget constraint is 50 = 10x + 5y. We can re-arrange the budget constraint as we did in Equation 7.4: y = y = 50 10 x 5 5 10 2x (7.5) Equation 7.5 is a line with an intercept at pm = 10 on the y-axis, an intercept of 5 on the xy axis and a slope of p = 2. Such a curve would look like bc1 in Figure 7.5. M - C H E C K From the budget constraint, we xpx know: y = pm py . We can take the first y derivative to see that the slope of the budget constraint is: dy px = dx py That is the opportunity cost of x in terms of y. 348 MICROECONOMICS - DRAFT Figure 7.5: Utility-maximizing consumption bundle. Harriet maximizes her utility subject to her budget constraint bc1 . At point a she consumes too little of x and too much of y (her marginal utility of y is much lower than her marginal utility of x, or her mrs(x, y) is too high, and she would be better off if she consumed less y and more x. Conversely, at c, she consumes too little of y and too much of x (her marginal utility of x is much lower than her marginal utility of y, or her mrs(x, y) is too low, and she would be better off if she consumed less x and more y. She maximizes her utility at b where her marginal rate of substitution, mrs(x, y) = uux , m y= py Gigabytes of data, y a mrs = mrt y yb equals her marginal rate of transformation or the p price ratio of x to y, mrt (x, y) = px . u3 b y u2 c u1 Budget constraint, bc1 xb x= Kilograms of coffee, x m px Checkpoint 7.1: Sketching a budget constraint a. Consider two goods: vegetables (x) which have a price of 4 euros per kilogram and meat (y), which has a price of 10 euros per kilogram. You have a budget of 50 euros a week for meat and vegetables for your family. Sketch your budget constraint. b. The price of vegetables increases to 5 euros per kilogram. What happens to your budget constraint? Sketch and explain. Budget constraints, indifference curves and the amount demanded In Figure 7.5 we show three of Harriet’s indifference curves. Remember at any given point on the indifference curve, the negative of its slope tells how much Harriet values the good on the x-axis compared to her valuation of the good on the y-axis, that is, her marginal rate of substitution (mrs(x, y)). Her marginal rate of substitution is her willingness to pay to get more of good x, namely how much of good y she would be willing to part with, in order to get one more unit of good x. So, (Negative of) the slope of an indifference curve = mrs = ux uy Harriet wants to get to the highest indifference curve that she can, given her budget. This is the point at which the budget constraint is tangent to her highest attainable indifference curve. For the two curves to be tangent, the slope of her indifference curve must equal the slope of the budget constraint. Or, R E M I N D E R We reason here in the same way we did in Chapter 3 that the utilitymaximizing choice is the point of tangency between the highest attainable indifference curve and the feasible frontier or budget constraint. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 349 the marginal rate of substitution must equal the marginal rate of transformation. mrs(x, y) = ux uy px = mrt (x, y) py = (7.6) This is the principle of constrained utility maximization that you learned in Chapter 3 because it determines how much of a good people will demand or want to buy at given prices. When Harriet uses the mrs = mrtrule, her trade-offs of one good for another in terms of utility (mrs(x, y)) equal the opportunity costs of the two goods in terms of each other (mrt (x, y)), where the opportunity costs are given by their prices. Remember that money she spends on one good means money she cannot spend on another good: capturing the essential idea of an opportunity cost. Equation 7.6 – the mrs = mrt rule – states that the relative valuation of x and y along her indifference curve must equal the opportunity cost of x in terms of y along her budget constraint. The point where Equation 7.6 holds, is Harriet’s utility-maximizing consumption bundle or the total quantity of goods x and y that Harriet will buy. A useful interpretation of the marginal rate of substitution occurs when good y is not data, but instead the amount of money left over from the budget after purchasing good x. In this case, the marginal rate of substitution can also be thought of as Harriet’s willingness to pay for the good x, the amount of money she will pay for an additional unit of x when she has already bought the quantity x. We need to distinguish between two concepts derived from the mrs = mrt rule: • Quantity demanded: The point (x, y) (the consumption bundle) at which p Harriet’s mrs(x, y) = px are her quantities demanded, these are the y amounts of each good that she consumes at the given prices and given income. • Demand function: Using the rule, we can derive a demand function for each good, where the quantity demanded depends on income (m) and prices ( px and py ). M-Note 7.2: The mrs = mrt rule applied to demand for goods The problem the person faces is to maximize their utility subject to their budget constraint: Vary x and y to maximize u(x, y) subject to the constraint y (7.7) = m py Substituting Equation 7.8 into Equation 7.7, the problem becomes: Vary x to maximize u(x, m py x px ) py px x py (7.8) W ILLINGNESS TO PAY A person’s willingness to pay for a good x in terms of y (for example budget left over to buy other goods) is their marginal rate of substitution between y and x when they are already purchasing the bundle (x, y). 350 MICROECONOMICS - DRAFT Taking the partial derivative of the utility function with respect to x, the first order condition for maximum utility is: ∂u ∂x ux ) uy mrs(x, y) px uy = 0 py = ux = px py = mrt (x, y) In the constrained utility maximum is a set of purchases such that the marginal rate of substitution is equal to the relative prices or the marginal rate of transformation. 7.2 Income, prices and offer curves To study how prices and incomes (budgets) affect the demand for goods we ask a hypothetical "what-if" question: how much of good x would someone purchase if her budget were m and the price of good x were px and the price of good y were py . A demand function shows the quantity purchased of x that results for the various values of the prices of both goods and the budget, px , py and m. So x(m, px , py ) is the demand for x as income (m) or the price of x ( px ) and the price of y( py ) change. We use the term demand curve when we refer to the simpler 2-dimensional graphical relationship x( px ) where we see how the amount purchased varies its price ( px ) varies holding constant all of the other influences on the demand for x. D EMAND FUNCTION , DEMAND CURVE A demand function provides an answer a hypothetical "what-if" question: how much of good x would a person purchase if her budget were m, price of goods x and y were px and py ? A demand curve is a 2dimensional graphical representation of a demand function showing the purchases of x that result for the various values of px (with the other influences on demand held constant). I NVERSE DEMAND FUNCTION , INVERSE DEMAND CURVE The We sometimes use a demand curve in which, instead of quantity sold depending on the price x = x( p), price depends on the quantity sold, p = f (x). This is called an inverse demand curve based on the inverse demand function, because it is the mathematical inverse of the conventional demand function. The inverse demand function contains exactly the same information and the demand function and the inverse demand curve looks identical to the conventional demand curve (it is downward-sloping). What differs is the hypothetical question for which the inverse demand function provides an answer. Instead of asking how much of a good will be purchased at a given set of prices and a budget, the inverse demand function answers the question: if the budget and the price of the other good are m and py what is the maximum price px that the buyer would be willing to pay to purchase an amount x of the good? A change in income: The income-offer curve To understand these changes, therefore, we examine Figures 7.6 a. and 7.6 b. As Figure 7.6 a. shows, as Harriet’s income changes, her budget constraint shifts. That is, the intercept with both axes, pm and pm , shifts up as income (m) y x inverse function answers the hypothetical question: what is the highest price that a person be willing to pay in order to purchase a given amount of some good, given her budget and the prices of other goods? The inverse demand curve is the simplified 2-dimensional graphical representation of this function as px = f (x). m3 py m3 py m2 py m2 py Gigabytes of data, y Gigabytes of data, y D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S m1 py bc1 bc2 u1 u3 u2 Income−offer curve c m1 py b a bc3 bc1 m1 px m2 px 351 m3 px x = m 1 px Kilograms of coffee, x bc2 x = m 2 px bc3 x = m 3 px Kilograms of coffee, x (a) Budget constraint & shifts in Income (b) The income-offer curve goes up and shifts down as her income goes down. Consider the three budget constraints in Figure 7.6 a where only income changes, but the prices of the two goods do not change. m m • Status quo: She starts with an income of m2 with intercepts p 2 and p 2 . y x • Income decrease: If her income decreases to m1 , then the intercepts of her m m budget constraint shift downwards and to the left to p 1 and p 1 , so she can y x buy less of both goods. • Income increase: If her income increases to m3 , then the intercepts of her m m budget constraint shift upwards and to the right to p 3 and p 3 , so she can y x by more of both goods. Considering different levels in Harriet’s income we can superimpose Harriet’s indifference curves to find the consumption bundle for each income level that would maximize Harriet’s utility using the mrs = mrt rule. The path traced out by the points (x, y) as m increases is called her income-offer curve. Her income-offer curve is also called her expansion path because it shows the effect of expanding her feasible set (by increasing her budget). In Figure 7.6 b, her income-offer curve is upward-sloping, showing the effect of an increase on her income on her consumption of the goods, x and y. As she gets more income, she would consume more of both goods. The income-offer curve allows us to understand the types of goods people consume. • Normal goods: normal goods are goods like coffee and data where people buy more as their income increases, or less of them as their income decreases. • Inferior goods: inferior goods are goods like cheap staples, like white Figure 7.6: Harriet’s budget constraint with shifts in income & her income-offer curve. In panel a Harriet’s budget constraints with three levels of income are shown (m1 , m2 and m3 ) with the corresponding budget constraints bc1 , bc2 and bc3 shifting outwards as income increases. In panel b Harriet’s budget constraints are shown tangent to three indifference curves, u1 , u2 , and u3 . The points where they are tangent are where mrs(x, y) = mrt (x, y). The curve joining all the points at which Harriet maximizes her utility as her income changes illustrate her income-offer curve. I NCOME - OFFER CURVE The income-offer curve is the path traced out by the points (x, y) that maximize the decision-maker’s utility as money-income, m, increases, holding the price of good x, px , and the price of good y, py constant. 352 MICROECONOMICS - DRAFT less of inferior goods as their income increases and more of them as their income decreases. Figure 7.7 shows a situation in which Harriet’s income increases, but her consumption responses for the two goods differs. For good y, on the vertical axis, Harriet consumes more of it as her income increases from m1 to m2 : she Quantity of normal good, y sandwich bread, basic rice, or instant noodles: people tend to consume h y3 g y2 bcm3 f y1 bcm1 increases her consumption from y1 to y2 . For good x, on the horizontal axis, on the contrary, Harriet consumes less as her income increases from m1 to m2 : she decreases her consumption from x2 to x1 as her income increases. As a result, she has a downward-sloping income-offer curve. Checkpoint 7.2: Inferior indifference curves 1. On your own set of axes, re-draw Figure 7.7. What condition must be true at each of points f, g and h? 2. Add the relevant indifference curves to your figure. What do you think they look like? Explain. A change in prices: the price-offer curve As prices increase, Harriet has to reduce her purchases of at least one good, and possibly both goods because she’ll have less money to spend on the goods if they become more costly. As the price of a good changes, holding the price of the other good constant, the slope of the budget constraint will change. For example, if the price of good x changes while the price of good y remains the same, the budget constraint will pivot around the y-axis intercept depending on whether the price of x increases or decreases. In Figure 7.8 a. on the left-hand side we show budget constraints based on three prices for good x: • An initial price along budget constraint bc2 . • A price increase, which steepens the slope of the budget constraint to bc1 . • A price decrease, which flattens the slope of the budget constraint to bc3 . At the point of tangency of each budget constraint with a corresponding indifference curve in panel b, we can see how Harriet would respond to a change in the price of goods x and y, other things equal. The path traced out by the number (x(m, px ), py ) as px changes (holding the budget, m and py constant) is called her price-offer curve. We illustrate the price-offer curve in Figure 7.8 b. Points a, b and c, all map her utility-maximizing choices and the curve that connects these points is the price-offer curve as we introduced in Chapter 3. bcm2 Income−offer curve x1 x2 x3 Quantity of inferior good, x Figure 7.7: Inferior Goods and An Increase in Income. The downward-sloping income-offer curve for an inferior good (x) and a normal good (y). As the person’s budget increases, they consume less of good x, showing that x is an inferior good (decreasing from x3 to x2 to x1 ). Consumption of good y increases as income increases, showing that y is a normal good (increasing from y1 to y2 to y3 ). D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S Price−offer curve m Money left over, y Money left over, y m = 10 bcpx3 px3 = 1 bcpx2 px2 = 0.5 m = 10 px3 c y3 u2 a y1 m = 40 px1 x1 = 6 u1 x2 = 9 x3 = 10.5 Kilograms of fish, x (a) Budget Constraint & changes in the price of x (b) The price-offer curve 7.3 Cobb-Douglas utility and demand You already encountered Cobb-Douglas utility in Chapter 3. We build on that base and explore a person’s choice of her utility-maximizing consumption bundle using indifference curves based on a Cobb-Douglas utility function and budget constraints. The Cobb-Douglas utility function has the following With a and 1 bc2 px = 0.5 bc1 px = 1 Kilograms of fish, x general form: u3 bc3 px = 0.25 b y2 bcpx1 px1 = 0.25 m = 20 px2 353 Figure 7.8: Harriet’s budget constraint with changes in price & her price-offer curve. In panel a Harriet’s budget constraints for three prices for fish, good x, px1 , px2 and px3 are shown with the corresponding budget constraints pivoting inwards for a price increase (such as for bc px2 to bc px3 ) or pivoting outwards for a price decrease (such as for bc px2 to bc px1 ). The other good, y, is money for other goods she can purchase. The intercept for good y is pm . Because y u(x, y) = xa y1 a (7.9) a indicating the relative strength of preference for the two goods and the intensities sum to 1. Using Cobb-Douglas utility, we can illustrate indifference curves for each good as the price of the good changes and we can derive a demand curve for each good. In Figure 7.9, we show the indifference curves for two goods: kilograms of coffee (x) and gigabytes of data (y). The marginal rate of substitution is: ux mrs(x, y) = uy = ✓ a 1 ◆⇣ ⌘ y a x (7.10) The mrs = mrt rule, then requires that, mrs(x, y) = ✓ a (1 ◆⇣ ⌘ y a) x = px = mrt (x, y) py (7.11) From this relationship and the budget constraint as we show in M-Note 7.4, we can derive a demand curve for each good. The demand function for kilograms of coffee, good x is: Demand x(m, px ) = am px (7.12) the price of money for other goods is py = 1, this simplifies to m as in Figure b. Note the x-axis range differs in panels a and b. In panel b Harriet’s budget constraints are shown tangent to three indifference curves, u1 , u2 , and u3 . The points where R E M I Nthe D Ebudget R Theconstraint marginaland rateindifference of substicurves tangent arefor where the marginal rate of tutionsare means that a given ratio of the substitution (mrs(x, y)) equals the marginal rate of quantities of the two goods x and y making transformation (mrt (x, y)). The curve joining all the up a bundle, much the person val-as the points at whichhow Harriet maximizes her utility price good xincrement changes isinher price-offer curve. ues aofsmall the amount of x compared to how much they value a small increment of y is given by the ratio of the exponents. For example, if a = 0.75, and x = y, then the marginal utility of x is three times the marginal utility of y (because a/(1 a ) = 3). 354 - DRAFT MICROECONOMICS Gigabytes of data, y m = 12 py a 6 b c Price−offer curve u3 x u2 bc1 px = 3 2 u1 bc3 px = 1 bc2 px = 1.5 4 6 8 12 p a Price of coffee, px p=3 b p = 1.5 c p=1 0 2 4 6 Demand curve for coffee 8 12 Kilograms of coffee, x Equation 7.12 shows a relationship between quantity demanded (x) and price ( px ) such that the quantity demanded decreases as the price increases, or the quantity demanded increases as the price decreases. Rearranging equation 7.12 as x·mx = a with a Cobb-Douglas utility function, a person implementing the mrs = mrt rule will spend a fraction of their total budget on x, that is: • equal to the exponent of x in the utility function namely, a a constant, and therefore is, • independent of the price of x and the price of y. The fraction of the budget spent on the other good is also independent of changes in the price of x. So, because the budget m has not changed, the amount spent on y will also remain the same. Because the price of the other good has not changed, the amount of good y purchased is also un- Figure 7.9: Harriet’s price-offer curve & demand curve for coffee. In the upper panel we show three of Harriet’s budget constraints corresponding to three different prices: p = 3 for bc1 , p = 1.5 for bc2 and p = 1 for bc3 . At points a, b, and c tangent to three indifference curves, u1 , u2 , and u3 The curve joining all the points at which Harriet maximizes her utility for different prices of good x is her price-offer curve. In the lower panel, we see how the use of the mrs = mrt rule in the top figure is the basis of the demand curve. The prices shown on the vertical axis are the (negative of the) slopes of the budget constraints in the top figure. We have assumed that a = 0.5, and that her budget for coffee and data is $100. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S changed. Thus, With Cobb-Douglas utility for a given price of y, py and income (m), the only thing that differs with different prices of x, is the quantity demanded of good x. This is why the price-offer curve in Figure 7.9 is horizontal. Similarly the amount spent on x does not depend on the price of y, which you can confirm from the fact that py does not appear in the demand function for x. These are not general features of demand functions, they are specific to the Cobb-Douglas utility function. M-Note 7.3: Marginal rate of substitution, Cobb-Douglas utility function Consider a Cobb-Douglas utility function: u(x, y) x a y1 = a The marginal utility with respect to of each good is: ∂u ∂x ∂u uy = ∂y ux = = axa = (1 1 1 a y a ) xa y a Therefore, the marginal rate of substitution of x with respect to y is: mrs(x, y) = ux uy = axa 1 y1 a ( 1 a ) xa y a (7.13) Note that: xa 1 xa y1 a y a and = 1 x = y Thus, the marginal rate of substitution (Equation 7.13) becomes: mrs(x, y) = ux uy = ✓ a (1 a) ◆ x y The inverse demand function We can re-arrange the function and find the inverse demand curve. The inverse demand curve is: Inverse demand = px (x, m) = am amount spent on x = = price x amount of x purchased Here, we have a downward-sloping demand curve where price decreases as the quantity demanded increases. We show this demand curve in the lower panel of Figure 7.9. Checkpoint 7.3: Changes to prices or income with Cobb-Douglas utility Harriet buys coffee and cookies to fuel herself while running her business. Her 355 356 MICROECONOMICS - DRAFT utility function for cookies (x) and cups of coffee (y) is given by the following utility function: u(x, y) = x0.6 y0.4 (7.14) We assume that Harriet has a weekly budget of $10 to spend on coffee and cookies, where the price of a cup of coffee is $3 and the price of a cookie is $0.50. a. Sketch Harriet’s indifference curves, her budget constraint, and calculate her utility-maximizing consumption bundles of cookies and coffee. b. Assume that the price of cookies now increases to $2 per cookie. What would happen to the quantity of cookies that Harriet demands? M-Note 7.4: Cobb-Douglas Demand Functions The Cobb-Douglas utility function is: u(x, y) = xa y1 a where 0 < a < 1. The individual maximizes this function subject to a budget constraint: m px x + py y = (7.15) Remember that the negative of the slope of the indifference curve is the marginal rate of substitution and the negative of the slope of the budget constraint (which is also the ratio of the prices of the two goods) is the the marginal rate of transformation. We found in M-Note 7.3 the following: mrs(x, y) ✓ = a 1 a ◆ y x So the utility-maximizing bundle that implements the mrs following equation: mrs(x, y) = ✓ a 1 ◆ y x = px py ) py y = px x a (7.16) mrt rule must satisfy the = ✓ a 1 a ◆ (7.17) To find the demand function, substitute Equation 7.17 into the budget constraint, Equation 7.15, to isolate a value for x, which we then use to find y: m = m = ) x(m, px , py ) = px x + px x (a + 1 Substitute px x into Equation 7.17 to find py y and y: py y = (1 a )m (1 a )m py y(m, px , py ) = a 1 a a ) px x a am px ✓ ◆ D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S We have therefore found the demands (x(m, px , py ), y(m, px , py )) as functions of the budget, m, and the prices of the goods, px and py , given the intensity of preferences for the goods, a . Notice that the demand for each good is independent of the price of the other good. The demand function for each good in terms of its own price is a hyperbola. Checkpoint 7.4: Income-offer and price-offer curves 1. Sketch the income-offer curve for Cobb-Douglas utility. Draw three CobbDouglas indifference curves with three income levels and sketch the corresponding income-offer curve. 2. The price-offer curve for Cobb-Douglas utility is a horizontal line in (x, y) space. Make sure that you can sketch a similar figure to the one we used in Figure 7.9. Notice, though, that in that figure a = 1 a = 0.5. In Chapter 3 we used a = 0.6 and 1 a = 0.4 to consider different preferences for Living (x) and Learning (y). How would these different exponents change the slopes of the indifference curves? Sketch approximately the corresponding price-offer curve for two goods, coffee (x) and cookies (y) for three different prices of coffee. 7.4 Application. Doing the best you can dividing your time We can apply the Cobb-Douglas utility function to a problem we all face: how to divide up the limited number of hours in our day between all of the things we would like to do, or must do to make a living. We simplify the problem by limiting our objectives to only two things: free time and consumption (similar to the problem involving Living and Learning in Chapter 3). Because we pay for our consumption with the wages we receive for working, and working means not having free time, we face a trade-off: more free time means less consumption (and of course vice versa). A trade-off between free time and consumption Consider a worker, Scott, deciding about how much leisure and consumption he would like. He is going to accept one job offer from the many job offers he is fortunate to have. In the jobs open to him, the hours of employment differ substantially as does the salary. We define h as the fraction of the day that Scott spends working for wages, with f = 1 h the fraction of that day that is free time ( f ). Scott consumes his entire income, so his daily consumption, x, is the total income he would receive if he worked 24 hours, w, times the faction of the day that he works. Consumption x = Wages ⇥ Proportion of day spent working = wh (7.18) Scott’s utility is given by the following Cobb-Douglas function that expresses 357 Leisure time as a proportion, f = 1 − h 358 MICROECONOMICS - DRAFT Figure 7.10: Feasible set and indifference curves for working time and consumption The choice of working time (and hence free time) and consumption. The worker chooses a level of consumption wh where their consumption is equal to their working hours (h) multiplied by their wage (w). The worker balances their consumption (x) against their leisure time as a proportion of their day. f=1 1 − ha a mrs = mrt u3 u2 = ua u1 xa = wha x=w Consumption, x = wh how he values consumption (x) and free time ( f = 1 u(x, h) = xa ( 1 h)1 h): a (7.19) Where, as before, 0 < a < 1 and the size of a indicates the relative intensity of preferences for the two goods. Because x = wh we can rewrite this as u = (wh)a (1 h)1 a (7.20) In Figure 7.12 we show Scott’s feasible set of choices between consumption and free time and three indifference curves representing Scott’s preferences for the two goods. The feasible frontier is your budget constraint. The maximum that Scott could spend on consumption is to have no free time and to set working time at 1 allowing a total expenditure of w on consumption. So w is analogous to m in the previous budget constraints and this limits expenditure on consumption x and the time not working, valued at the wages Scott would have received had Scott worked the entire day: Budget constraint: w x + (1 h)w (7.21) We can re-arrange Equation 7.21 as follows: Re-arranged constraint x wh (7.22) Equation 7.22 therefore requires that the value of Scott’s consumption x not be greater than wages for time worked wh. M - C H E C K We can rewrite the budget constraint as 1 h = 1 x w From which we see that d (1 h) dx = 1 w Which is the effect of greater consumption on feasible free time. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S The negative of the slope of the budget constraint is the marginal rate of transformation of reduced consumption into free time. This is how much additional free time Scott is able to have by giving up one unit of consumption. The equation for the budget constraint (Equation 7.21) and Figure 7.10 show that the marginal rate of transformation is as follows: D (1 h) = mrt (x, 1 Dx h) 1 w = (7.23) The slope of Scott’s indifference curves is the marginal rate of substitution between consumption and free time. As shown in Equation 7.16 the marginal rate of substitution derived from a Cobb-Douglas utility function is the ratio of the x exponent to the y exponent times the ratio of the value of the y variable to the x variable or: mrs(x, 1 h) ✓ = ◆⇣ ⌘ y a x a 1 The y variable here is the amount of free time 1 h and x is the level of con- sumption which is x = wh, so mrs(x, 1 h) = ✓ a a 1 ◆✓ 1 h x ◆ The best Scott can do in this constrained optimization problem, maximizing his utility subject to his budget constraint, is to select the bundle (x, 1 h) such that the marginal rate of substitution is equated to the marginal rate of transformation. At his utility-maximizing bundle the fraction of the day Scott will work, h, is equal to a (see M-Note 7.6). As a result, the fraction of Scott’s day that is free time will be 1 a , because 1 a , the exponent of "free time" in your utility function is a measure of how important free time is to you. M-Note 7.5: Consumption, free time, and work hours As explained in the previous section, we have: u(x, h) Scott’s utility function wh Scott’s budget constraint h) = mrs(x, 1 h) = using y = 1 MRT (using equation 7.21) mrt (x, 1 h) = xa ( 1 x (1 re-arranged and as an equality MRS (using equation 7.16) = h = D (1 h) Dx = h)1 a x 1 w ✓ ◆⇣ ⌘ a y 1 a x ✓ ◆✓ ◆ a 1 h 1 a x 1 w 359 360 MICROECONOMICS - DRAFT Country Average annual work hours of production workers Average annual work hours of production workers Country France Germany Netherlands Sweden Switzerland United Kingdom 3000 2500 2000 1500 Australia Canada Japan United States 3000 2500 2000 1500 1900 1913 1929 1938 1950 1960 1973 1980 1990 2000 1900 1913 1929 1938 1950 1960 1973 1980 1990 2000 Year Year (a) Work hours over time in European Countries To calculate Scott’s time worked, we equate the mrs(x, 1 mrs Using x = wh ✓ ✓ a a 1 a 1 a ◆✓ ◆✓ 1 h x 1 h wh wa (1 a ◆ ◆ (b) Work hours over time in countries outside of Europe h) to the mrt (x, 1 = 1 w = 1 w h) = (1 ah = h a = h h): Figure 7.11: Work hours over time for a variety of countries. The data refer to annual average work hours for full-time production workers (meaning, excluding supervisory personnel) Source: Huberman (2004) mrt a )wh ah (7.24) Scott’s hours worked only depends on a , or how much Scott values consumption relative to free time. Checkpoint 7.5: Preferences for free time and consumption of goods 1. 2. 7.5 Application: Social comparisons, work hours and consumption as a social activity In Chapter 3 we looked at data on how men and women spend their time, and the increase in the amount of time women spend working for pay (the female labor force participation rate). Here we use the constrained optimization model to help understand another dramatic change in time use over the last century. Figure ?? shows that in every country on which we have data people have D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 361 been working less. But there are important differences among the countries: • In the Netherlands work hours fell from the equivalent of 62 hours 52 weeks of the year to less than 27 hours per week. • In Sweden, where work hours also declined dramatically, there was a small increase in work hours from 1980 to 2000. • Work hours declined much less in the U.S. than in most other countries – a decline of 32.77 percent compared to a decline of 58.04 percent in the Netherlands. • In the US, as in Sweden there was a slight increase in work hours at the end of the last century. How can our model help explain what explains these differences? We modify the model of choice of work hours to help us understand the differences among the countries and the changes over time shown in Figure 7.13. The new idea is that what people consume - the quality, quantity and expense of what someone wears, or drives, or eats - is a signal to others and to themselves about where they stand in society relative to other people. That is, people judge their own level of consumption relative to other people’s consumption, not based on the level of consumption alone. Veblen effects, conspicuous consumption and working time A particular variant of this view of consumption as social signaling terms the things we purchase in order to impress ourselves or others conspicuous consumption and it is the high income people who set the standards for everyone else. One way to model this is to say that we compare our consumption to that of the very rich, and the closer our consumption is to the consumption of the rich, the better we feel. H I S TO RY The term "conspicuous consumption" comes from the American economist and sociologist Thorsten Veblen (18571929). Well over a century ago he described exactly the trade off we are modeling here. "The means of communication and the mobility of the population now expose the person to the observation of many persons who have no other means of judging his reputability than the display of goods... the present trend of the development is in the direction of heightening the utility of conspicuous consumption as compared with leisure."1 To do this we now define "effective consumption" as how our consumption feels to us given what others are consuming. To capture this idea, we define effective consumption as follows: Effective consumption x v = Consumption = x vx Veblen Effect ⇥ Consumption of the Rich Where, x is Scott the worker’s consumption, x is the consumption of the rich, and v is a positive constant representing the Veblen effect. The negative effect of the consumption of the rich on our utility is captured in the term v (named after Thorstein Veblen). Effective consumption defined in this way expresses the idea that the consumption of the rich has the effect of diminishing the adequacy that we feel for any particular level of consumption. (7.25) Leisure time as a proportion, f = 1 − h 362 MICROECONOMICS - DRAFT Figure 7.12: Feasible set and indifference curves for working time and consumption: Veblen effect. The worker now experiences a "Veblen effect" which changes their preferences over work time and leisure, altering the slope of their indifference curves and therefore resulting in a new utility-maximizing point. At the new utility maximum, they consume more goods, work more and consume less leisure. f=1 1 − ha a b 1 − hv ua uv2 uv1 xv = whv xa = wha x=w Consumption, x = wh This raised the marginal utility of greater consumption to compensate. Scott’s utility now includes this idea: u = (x vx)a (1 h)1 Using equation 7.16 for the mrs(x, 1 a = (wh vx)a (1 h)1 a (7.26) h) with a Cobb-Douglas utility function, Scott’s marginal rate of substitution is now: mrs(x, 1 h) = = ✓ ✓ a a 1 a 1 a ◆✓ ◆✓ 1 h ◆ xv ◆ 1 h x vx (7.27) Because increased consumption by the rich ((x)) has diminished your level of effective consumption, it has raised the mrs(x, 1 h) (you can see this in the second term in Equation 7.27). So how much Scott values consumption relative to how much he values free time is now greater than before. This means that in Figure 7.12 it has made the indifference curves steeper. Figure 7.10 shows an initial choice that the worker faces without the Veblen effect, that is, when he does not worry about what others consume. Figure 7.12, as a contrast, shows how the "Veblen effect" of the consumption of the rich affects the worker’s choice. The Veblen effect does not alter the feasible frontier, but it changes the marginal rate of substitution between consumption and free time. The indifference curves are steeper because at any level of actual (not effective) consumption and free time, the marginal utility of effective H I S TO RY The people who set consumption standards, according to Veblen, are the rich. He wrote: "all canons and reputability and decency and all standards of consumption are traced back... to the usages and thoughts of the highest social and pecuniary class, the wealthy leisure class."2 D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 363 consumption has risen. Why? Because when Scott compares himself to the rich it makes him feel as though he has less effective consumption than he actually has. Figure 7.12 shows that when Scott maximizes utility by setting the mrt (x, 1 h) to the mrs(x, 1 h) Scott works longer hours, enjoys less free time, and increases the amount of consumption he purchases (see the M-Note 7.6). The time Scott works now is greater than before, as shown in Figure 7.10 (see the M-Note 7.6): hv = a+ (1 a )vx w (7.28) Equation 7.28 shows the working time without the Veblen effect (namely, a ) plus more time that Scott is now motivated to work due to the Veblen effect (namely, the second term on the right hand side of Equation 7.28). Did Veblen effects and falling inequality explain declining work hours in the 20th century? How does this model help us understand how working hours have changed over time and how work hours differ across countries? The model predicts that the more that rich people consume, the longer other people will work. So we would expect people to work longer hours in countries in which the rich are especially rich, and people to work less where the rich are only modestly richer than the rest. Figure 7.13, presents the average annual working hours and a measure of the fraction of all income received by the richest 1 percent of households in ten countries over the 20st century. The fraction of income received by the very rich is a measure of the key variable in the model, the consumption of the very x rich divided by the wages of typical workers or w . The figure shows that this prediction of the model with Veblen effects is borne out by the data: longer working time is associated with a larger share of income going to the very rich. But it shows more: the decline in the relative incomes of the very rich is closely associated with the decline in work hours. Notice that Sweden is both the most unequal and "longest working" nation (in the early years of the data set) and also the most equal and country with the most free time (in the more recent years). The countries that became more equal over this period also saw the greatest drop in work hours. The increase in work hours in both Sweden and the U.S. at the end of the last century was associated with an increase in inequality in both countries. In this model, conspicuous consumption by the very rich is a kind of "public bad." It is experienced equally by all members of the society or at least can F AC T C H E C K In 2001 the tax authorities in Norway began posting income tax records online, so that anyone could find out the income of their neighbors, friends and co workers. Huge numbers of people accessed the site. Ricardo Perez-Truglia studied the statistical relationship between Norwegian’s income and measures of their "subjective well-being" – happiness and life satisfaction. Higher income people were happier and had greater life satisfaction. But after incomes became public the differences between rich and poor in subjective well-being became much greater.3 364 MICROECONOMICS - DRAFT Sweden, 1900 ● Country Annual Average Work Hours 3000 ● ● Netherlands, 1913 ●● ● ● ● ● 2500 ● ● ●● ● ● ● ● 2000 ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● Sweden, 2000 1500 ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● Canada ● France ● Germany ● Japan ● Netherlands ● Sweden ● Switzerland ● United Kingdom ● United States ●● ● ● Australia ● ● ● ● ● ● ● ● Netherlands, 2000 1.5 2.0 2.5 3.0 Income Inequality be by anyone with access to TV and social media (it is non-excludable). It is a "bad" and not a good because it reduces the utility of those it affects (it is non-rival in the disutility it creates). And the people affected then respond in ways that generate further negative external effects because the Veblen effect induces them to work and consume more, increasing the use of our limited environmental resources. The Veblen effects model suggests some of the reasons for differing working hours among countries. But by itself it misses some important parts of the story. The most important missing element for the decline in work hours in the 20th century is that voting rights were extended to include most adults early in the century. When overworked employees got the right to vote, in virtually all countries their trade unions and political parties demanded reductions in working hours. The model with Veblen effects and the data provide an illustration of a more general point: consumption is not just a biological activity. Eating is not just nutrition. Clothing is not just keeping warm. Your home is more than four walls to keep out the weather. Consumption is a social activity. As Veblen said our consumption is a signal to others and to ourselves about who we are. It is also a social activity in which we engage, for example, for the pleasure of the company of our friends. Figure 7.13: Inequality and work hours 1900 to 2000. The hours data are annual average work hours for full-time production workers. The income data are based on the share of total income received by the top 1 percent of households. Source: Oh, Park, and Bowles (2012). D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 365 M-Note 7.6: The choice of work hours with & without a Veblen effect Using equation 7.26, Scott’s utility function when deciding between leisure and consumption may be represented as: u = (x vx)a (1 h)1 a With v 0. When v = 0, there is no Veblen effect, and we have same utility function as in Equation 7.19 that we used in the previous section. When v > 0, there is a Veblen effect. To calculate the time worked, we need to equate the mrs(x, 1 Following 7.27 and 7.23: a mrs Using x = wh 1 a 1 h) to the mrt (x, 1 (1 h) a (x vx) = 1 w (1 h) a (wh vx) = 1 w h) = (1 a )(wh awh = wh awh wh = aw + (1 hv = a+ aw(1 aw h). mrt (1 vx) (1 a )vx a )vx a )vx w In the absence of Veblen effect (v = 0), the hours worked h depends only on the importance of consumption relative to free time in the utility function, a . A positive Veblen effect (v > 0) reduces effective consumption. Because there are diminishing returns to effective consumption (marginal utility of effective consumption is lower the more of it you have) the effect of there being less effective consumption is to increase the marginal utility of it. As a consequence, Scott increases his working hours (and consumption) and reduces his leisure. You can also see that the higher the consumption of the rich, the lower is effective consumption, and therefore the higher the hours worked (See Figure 7.12). Checkpoint 7.6: Veblen effects • What do you think explains the magnitude of v, the parameter governing the size of the Veblen effect? • Juliet Schor, an economist, found that people who watch TV more save less. Saving is not included in our model, but how do you think this result might be explained by some kind of Veblen effect?4 • How could our model explain the evidence in the Fact Check about the effects of making incomes public on the relationship between income and subjective well being? How would that affect v the Veblen effect coefficient? 7.6 Quasi-linear utility and demand ?? We don’t always want to know how people trade off the benefits of two commodities – like coffee and data – in their utility functions. Sometimes, we find it useful to think about how a person will trade off money left over for other purchases (y) and a commodity (x), like we saw earlier in the example of kilograms of fish (x) and money for other goods (y). We explore this idea using Q UASI - LINEAR FUNCTION When a function is quasi-linear it depends linearly on one variable, e.g. y, and non-linearly on another variable, e.g. x, and has the form u(x, y) = y + g(x). Hence it is quasi or "partly" linear. 366 MICROECONOMICS - DRAFT Figure 7.14: Harriet’s indifference curves: quasilinear utility. With quasi-linear utility, marginal rates of substitution depend only on the amount of the good x, and not at all on the amount of money left over to buy other goods, y. As a result indifference curves with different levels of utility are vertical displacements of a single curve. This means that the slopes of the indifference curves when consuming x0 amount of fish are the same independently of the amount of money she has for other purchases. y3 = 17 Money left over, y y2 = 15 y1 = 13 u3 u2 u1 x0 x1 Kilograms of fish, x a Quasi-linear (QL) utility functions with the form: u(x, y) = y + g(x) (7.29) Utility in the quasi-linear function depends linearly on y. The more y Harriet has, the higher her utility. Her marginal utility for y is always 1 and it does not decline as she gets more y. These properties make y more like wealth measured in terms of money than like a particular good such as data or coffee, so we will often refer to y as money, understanding that it is really generalized purchasing power that can be spent on many other things possibly in many periods. The analysis of demand is greatly simplified if the marginal rate of substitution depends only on the amount of the good someone purchases, and not on the amount of money she has left over. Here are the simplifications: • Prices: When y is money left over for other purchases, then py = 1 (the p price of a dollar is a dollar), so we can simplify mrt (x, y) = px = px = p. y That is, when the other good is money, we shall simply refer to the price of good x as p, which is the opportunity cost of x. • Marginal utility : when utility is quasi-linear in y, then uy = 1, that is, the marginal utility of money is constant and equal to 1. Therefore the marginal rate of substitution mrs(x, y) = uux = ux is just the marginal utility of x, which y is the trade-off of money for x. Q UADRATIC, QUASI - LINEAR UTILITY In the case of quadratic, quasi-linear utility, the non-linear part of the utility function, g(x) is a quadratic function of x such as g(x) = p̄x 12 ( p̄/x̄) x2 , where x̄ is the maximum amount of x someone is willing to consume and p̄ is their maximum willingness to pay for x when they have not yet consumed any x. The quadratic quasi-linear utility is quadratic in x (it has a "squared" term in x) and linear in y. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 367 • Willingness to pay : Because with the quasi linear utility function the mrs is simply ux (x, y), that is, the marginal utility of x when consuming the bundle (x, y), the mrs is also the maximum amount (in money units) that the person would be willing to pay to have a small increase in the amount of x. If we use a quasi-linear utility function in which money left over is the linear good, then utility is measured in money. To see this suppose the person spent nothing on the other good; then Equation 7.29 tells us that u = y. Utility is the amount of money the person has to spend. It is also the case that if the amount of money the person has to spend increases by $10, then utility increases by 10. So the units in which utility is measured is money. This does not mean that the only thing the person cares about is money. Whatever x is may matter a lot. It is just that how much it matters will be measured in money equivalents. Quadratic, quasi-linear utility p̄, x̄, SATIATION & BLISS Many of the examples in this book use the particular class of quadratic quasilinear utilities, where the function g(x) = p̄x 1 2 p̄ x̄ x2 is a quadratic function of the good x, and as a result the utility function is: Quadratic, quasi-linear utility u(x, y) = y + p̄x pay for good x. She won’t pay more than p̄ to get a unit of x. x̄ is the person’s satiation point for x, ✓ ◆ 1 p̄ 2 x 2 x̄ (7.30) In Equation 7.30: • Satiation: The parameter x̄ represents the level of x at which the buyer is satiated with x and would consume no more even if the price were zero. • Maximum willingness to pay: The parameter p̄ > 0 represents the buyer’s maximum willingness to pay for the first unit of x when they do not have any x, i.e. when x = 0. The marginal utility of x with the QQL utility is: Du = ux (x, y) = p̄ Dx p̄ is the person’s maximum willingness to p̄ x x̄ (7.31) Equation 7.31 tells us the following: • When x < x̄, the buyer’s marginal utility of x is positive, and she regards x as a good. • When x > x̄, the buyer’s marginal utility of x is negative, and she regards x as a bad. If y is budget left over to buy other goods, then the marginal utility of y is always 1, regardless of the levels of x and y. As a result, the marginal rate of beyond which her marginal utility of x is negative. She would prefer not to consume x > x̄. The point at which you are sated (verb) is where you reach satiation (noun) from consuming a good, like x. The intuition is easily seen with food: you reach satiation at that point where you do not want to eat another mouthful (the marginal utility hits zero) or, if you do, you know you’ll regret it (the marginal utility will be negative). Or, it is the point at which you have reached bliss, which is perfect happiness or great joy, and at which, if you consumed or did any more, it would detract from that bliss, joy or wonder. 368 MICROECONOMICS p - DRAFT Figure 7.15: Harriet’s marginal rate of substitution (demand): quadratic quasi-linear preferences. The figure shows Harriet’s mrs(x, y) for a good, x, and money for other goods, y. That is, the downward-sloping line is Harriet’s willingness to pay in money for an additional unit of good x for different levels of the quantity she has of good x changes and is therefore also her demand curve because it shows the relationship between the price of the good and how much of it she will buy at different prices. The vertical intercept, p̄, is the person’s maximum willingness to pay when they currently consume zero units of good x. The horizontal intercept, x̄, is the person’s satiation point or bliss point beyond which Harriet’s marginal rate of substitution is negative so that x changes from being a good to a bad. Price per unit of good x, p p = Maximum Willingness to Pay ⎛p⎞ mrs(x, y) = p − ⎜ ⎟x ⎝x⎠ x = Point of Satiation Quantity of the good, x x substitution is equal to the marginal utility of x: mrs(x, y) = ux = ux = p̄ uy p̄ x x̄ (7.32) We can think about Equation 7.32 in the following way: • Equation 7.32 is the equation for a line. • Equation 7.32 has vertical intercept p̄, which is the buyer’s maximum willingness to pay. • Equation 7.32 has a horizontal intercept x̄, which is the point beyond which the buyer does not want to pay for good x (at x = x̄, the buyer’s willingness to pay is zero). x̄ is the buyer’s bliss point. To get the buyer to consume more than x̄ of the good, you would have to pay her, rather than expecting her to pay you. • Equation 7.32 shows that the marginal rate of substitution has a slope of p̄ x̄ , which is the negative of the ratio of the maximum willingness to pay to the satiation point. Equation 7.30 shows that Harriet’s utility for the good x depends on how much she has relative to a target level x̄. When 0 x x̄, Harriet’s utility increases if she has more x. But when x x̄, Harriet’s utility decreases as she gets more x. We can plot quadratic, quasi-linear marginal rate of substitution or willingness-to-pay as a function of x as in Figure 7.15. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 369 The inverse demand function, which gives the highest price Harriet will pay for each given total amount of the good is: p(x, m) = p̄ Inverse demand function: p̄ x x̄ (7.33) We will often simplify the inverse demand function such that the slope of the p̄ function, x̄ is represented by a simple slope coefficient, b . We will therefore represent the demand curve as follows: Demand curve: p(x, m) = p̄ bx (7.34) Suppose Harriet has quadratic quasi-linear preferences between a good x and money left over to buy other things, with the willingness-to-pay p(x) = mrs(x, y) = p̄ p̄ x̄ x. She starts out with budget m and has the opportunity to buy any amount of the good x at the price p. If p(0) = p̄ > p, Harriet will buy at least some of the good. If she buys x0 units of the good, Harriet’s willingness-to-pay will have fallen to p(x0 ). If p(x0 ) > p, then Harriet will buy more of the good, and so on, until p(x) = p. At this point Harriet doesn’t buy any more of the good, and spends the rest of her budget on other things. So from Harriet’s willingness-to-pay (or marginal rate of substitution), we can derive the quantity demanded at a market price p, which tells us how much of the good she will buy when the market price for the good is p. x( p) = x̄ x̄ p p̄ (7.35) We can also re-write Equation 7.33 as a demand function Demand function x(m, p) = x̄ x̄ p p̄ (7.36) Equation 7.36 says that for p > 0 a person will consume an amount less than their point of satiation (x̄) given the price of the good p and their maximum willingness to pay p̄. As the price of x increases, then the quantity demanded of x will decrease. Figure 7.16 demonstrates this relationship by showing Harriet’s utility maximizing choices between x and y with her indifference curves and budget constraints for three prices of x in the top panel, while also showing her marginal rate of substitution of money for the good in the lower panel. The lower panel also shows how Harriet’s marginal rate of substitution corresponds to a demand curve, by showing three different price levels and how the given price determines the quantity demanded at that price. M - C H E C K For example, when the market price is: p = $10, p̄ = $20, and x̄ = 10, then Harriet would like to buy 5 apples since her willingness-to-pay is the following: mrs(5, y) = $10 for any y. Money left over, y 370 - DRAFT MICROECONOMICS Figure 7.16: Harriet’s utility-maximizing choice and marginal rate of substitution (demand): quasi-linear preferences. The top figure shows Harriet’s indifference curves for kilograms of fish (x) and money left over for other goods (y). The figure shows her utility-maximizing choices at three levels of prices for a kilogram of fish. Harriet’s 2 utility function is u(x, y) = y + 20x 12 20 10 x . Her budget constraint is y = 600 px x. The lower panel shows Harriet’s mrs(x, y) for a good, x, and money for other goods, y. Harriet’s marginal rate of substitution is therefore mrs(x, y) = 20 2x. Her marginal rate of substitution, as her willingness to pay in money (y) for goods (x) is her demand function for x. She has a y-intercept of y = 20 = p̄ (her maximum willingness to pay) and her xintercept is x = x̄ = 10 (the amount of fish that satiates her appetite for fish, which is also the maximum quantity of fish she would consume were the price of fish zero). The slope of her marginal rate of substitution suggests she will exchange money (y) for fish (x) until her mrs(x, y) = px , i.e. when 20 2x = 10, which implies x = 5 when px = 10. 600 u3 558 a c 550 b u2 bc3 pL = 6 u1 bc2 pB = 10 bc1 pH = 14 xH = 3 Price per unit of the good, p p = 20 pH xB = 5 xL = 7 x = 10 mrs(x, y) = 20 − 2x High price, pH = 14 f Base price, pB = 10 e pB g pL xH = 3 xB = 5 Low price, pL = 6 xL = 7 Quantity of the good, x x = 10 M-Note 7.7: The demand for x and y with with quadratic quasi-linear preferences We begin with the mrs = mrt rule, and then rearrange it to give us the demand for good x mrs(x, y) = p̄ p̄ x x̄ p̄ x = p = mrt (x, y) x̄ = p̄ p x(m, p) = x̄ x̄ p p̄ (7.37) The person will then use whatever remains of their budget (m) as money to spend on D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 371 other goods, y, given what they spent on x at its price, p: y = m y = m y(m, p) = m px ✓ p x̄ ◆ x̄ p p̄ 2 x̄ px̄ + p p̄ (7.38) Equation 7.38 shows that once we determined the demand for x, we can derive the demand for the other goods as well. 7.7 Price changes: income and substitution effects When the price of a good changes, the consumption of the goods changes as we saw when deriving the demand curve from the offer curve in Figure 3.13. The total amount of the change is made up of two components: • The income effect • The substitution effect The income effect occurs because a person has more or less purchasing power as a consequence of the change in prices. A person’s real budget is a measure of their purchasing power with a given budget – the maximum amount they can purchase of a good at given prices. If the price of the good I NCOME AND SUBSTITUTION EFFECTSWhen the price of a good changes there are two effects. The first, due to the change in relative prices (the good’s price relative to other goods’ prices), leads people to substitute among the goods they buy. This is the substitution effect. The second effect arises because the price change alters peoples’ real income, expanding or shrinking the feasible set of purchases. This is the income effect. The two effects are calculated so that they sum to the total change in the quantity of the good consumed. increases the consumer can buy less of the good than before the price increase. They have less purchasing power at their fixed budget. Generalizing this to two goods, if the price of one good decreases while the price of the other good remains the same, the person can buy more of both goods when the price of one good decreases. The substitution effect occurs when, as the prices of goods change, people substitute away from goods that are relatively expensive and they will substitute towards goods that are relatively cheap. So, when choosing between two goods that both provide positive utility, when the price of one increases relative to the other good, the person will normally substitute away from that good toward the other good (assuming some degree of substitutability between the goods). Income and substitution effects for normal goods Consider a change in the price of good x, for example, an increase in px . Harriet will change the amount of good x that she buys. We can decompose or separate this change into a substitution effect, which represents how Harriet would change her choice if she had to move along the original indifference curve, and an income effect, which represents the change in her behavior due to moving to another (in this case, lower) indifference curve. Harriet starts at point a with bundle (xa , ba ) on her initial budget constraint bca before the price C OMPENSATED BUDGET CONSTRAINT The compensated budget constraint takes the new prices of goods as given (it is parallel to the budget constraint after the price change), but it gives the person sufficient income to return to their original indifference curve, therefore creating a new point of tangency with the original indifference curve. In the case of a price decrease a compensated budget constraint would take money away from a person, for example via a lump sum tax. 372 MICROECONOMICS - DRAFT Gigabytes of data, y u1 c yc yb = ya Figure 7.17: Income and substitution effects: Cobb-Douglass utility. The substitution effect is shown by the hypothetical movement along u2 from a to c. The income effect is shown by the movement to another indifference curve, from b to c with the income effect being the difference in the purchases of coffee between those two points. The fact that the change in the price of coffee does not affect the level of purchases of data ya = yb is an attribute of the specific Cobb-Douglas utility function we have used here, it is not a general result. u2 a b bca initial budget bcb budget when px increases xb xc Income effect bcc compensated budget xa Substitution effect Kilograms of coffee, x increases. When the price increases, the budget constraint pivots inwards to bcb and creates a new equilibrium at b with bundle (xb , yb ). The income effect can change both the amount of good x and good y, as Figure 7.17 shows. To decompose this change into an income and a substitution effect, we need to construct a counter-factual. We take Harriet back to her original indifference curve, but we retain the slope of the new budget constraint to see how the new prices would have affected her. To break down the effects, we use the idea of a compensated budget constraint. The compensated budget constraint could hypothetically be implemented by a policy-maker who wants to ensure that a person who has lost real value of their available budget because of price increases can be compensated in some way, for example, by a government transfer. The compensated budget constraint bcc allows us to consider how Harriet would respond to the new price for good x if she were limited by the compensated budget constraint that provides Harriet with just enough money to be on her original indifference curve, u2 . The new tangency is at point c at bundle (xc , yc ). The difference between xa and xb is the total effect of the price change or the substitution effect plus the income effect. By construction the substitution effect causes a movement along the original indifference curve to point c. The E FFECTS OF A PRICE CHANGE The total effect of a price change is the change in quantity demanded. The decomposed effects shows how the total effect is broken up (decomposed) into the two parts of the substitution effect (a movement along the indifference curve) and the income effect (a movement to a new indifference curve). D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S Utility Function Income Effect Substitution Effect Cobb-Douglas Yes Yes Quadratic, quasi-linear No Yes Perfect Complements Yes No 373 Table 7.1: Utility Functions and their income and substitutions effects. Remember that the substitution effect is captured by a movement along an indifference curve as prices or real incomes change, whereas the income effect is captured by a movement to a new indifference curve. difference between xc and xa is the substitution effect. The difference between xb and xc is the income effect, which is what drives Harriet’s choice to point b with bundle (xb , yb ). <Insert links to online appendices with no income effect and no substitution effect>. Complements and substitutes in consumption The size of the income and substitution effect will depend on whether the goods "go together" or have an "either/or" quality. If two goods are more enjoyable consumed together, then they are complements (coffee and cookies). "Either/or" goods are called substitutes: they are consumed instead of R E M I N D E R In Chapter 6 we introduced the idea that in inputs to a production process may be either substitutes (computer driven machine tools and skilled machine operators) or complements (computer driven machine tools and engineer-programmers). each other (tea and coffee). At the extreme, perfect substitutes have a constant marginal rate of substitution (linear indifference curve), whereas perfect complements (right shoes and left shoes) are only valuable when consumed together. Indifference curves for perfect complements are L-shaped. M-Note 7.8: Complements and substitutes in consumption Complements: Cookies and coffee. In the Cobb-Douglas utility function u (coffee) and y (tea) are complements because ∂u = axa ∂x = xa y1 a, x 1 1 a y which, because 0 < a < 1, means that ∂ u/∂ x = (1 ∂y a )axa 1 y a = (1 x a )a ( )a > 0 y so the greater is the consumption of y the higher is the marginal utility of x. By the same reasoning, the greater is the consumption x, the higher is the marginal utility of y. Substitutes: Coffee and tea. Here is a utility function in which x (coffee) and y (cookies) are substitutes: u = (x + ey)a where e is a positive constant measuring how much the person prefers tea to coffee and 0 < a < 1. Therefore, finding the marginal utility of x by taking partial derivatives: ∂u = a (x + ey)a ∂x 1 >0 The marginal utility of x is positive. But how does the marginal utility of x change as consumption of y changes? We can work that out by taking the partial derivative of the marginal utility of x with respect to y: ∂ u/∂ x = e (a ∂y 1)a (x + ey)a 2 <0 It is negative because a < 1. This shows that the marginal utility of coffee is less the more tea the person consumes. The same reasoning shows that the marginal utility of tea is C OMPLEMENTS AND SUBSTITUTES IN CONSUMPTION Goods are complements in consumption if an increase in the quantity consumed of one raises the marginal utility of the other. Goods are substitutes in consumption of an increase in the quantity consumed of one reduces the marginal utility of the other. The property of being complements or substitutes is symmetrical: If good x is a complement of good y, then y is also a complement of good x. The same is true for substitutes. 374 MICROECONOMICS - DRAFT less, the more coffee the person consumes. For this particular utility function, tea and coffee are what is called perfect substitutes. In this case, ∂u = ea (x + ey)a ∂y 1 so the individual’s marginal rate of substitution, the ratio of the two marginal utilities, is as follows: ∂ u/∂ x a (x + ey)a = ∂ u/∂ y ea (x + ey)a 1 1 = 1 = mrs e Because the mrs(x, y) is the negative of the slope of an indifference curve, the fact that it is constant means that the indifference curves are linear. This is what the fact that x and y are perfect substitutes means. 7.8 Application: Income and substitution effects of a carbon tax and citizen dividend The decomposition of the results of price changes into income and substitution effects can be illustrated by the proposed carbon tax to reduce emissions of carbon dioxide and other greenhouse gases that contribute to climate change. The prices of petroleum, coal, natural gas and other fossil fuels do not include the costs of the environmental and climate-change external effects of their use . This means that people pay a private cost of using fossil fuels that is lower than the social (private plus external ) cost of using them. The result – as in the case of over-fishing in Chapters 1 and 5 – is overuse of fossil fuels. We now consider tax imposed on the sale of fossil fuels, in conjunction with a transfer of the tax resulting revenues in equal amounts back to the members of the population, called a citizen’s dividend. We ask: how would this so-called carbon tax and citizen dividend policy reduce consumption of fossil fuels and affect citizens’ consumption of other goods. We consider two steps in our policy process: • The substitution and income effects of the increased price of fossil fuels. • The income effect of the citizen dividend. Reducing carbon emissions by imposing the tax In Figure 7.19, the fossil fuel consumption (x) of a citizen is plotted on the horizontal axis, and the consumption of others goods measured in some currency is plotted on the vertical axis (y). Before the tax, a citizen is at point a in Figure 7.19 where the marginal rate of substitution equals the marginal rate of transformation, which is the existing price of fossil fuels, pa . At a, the citizen has a utility of u2 on the corresponding indifference curve. The government then imposes a tax on fossil fuels. The tax increases the Figure 7.18: A conservative case for climate action. In their 2017 op-ed (opinion editorial) in The New York Times three economists made "A conservative case for climate action," proposing the carbon tax and citizen dividend that we analyse here. They included Martin Feldstein and Gregory Mankiw, chief economic advisors to U.S. Presidents Ronald Reagan and George W. Bush, respectively. R E M I N D E R In Chapter 5 we explained that the social cost equals the private cost plus the (negative) external cost imposed by a person’s action. In that case, a person’s marginal private cost of fishing included their disutility of fishing, but the social cost included not only the private costs but also the negative external effect the fishermen imposed on others. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S md mb = ma c yc yb = ya a b bcb budget when px increases bcc compensated budget bca initial budget u1 mb d yd b yb xb effect Fossil fuels consumed, x (a) Substitution and income effects of the price increase a bcd budget with dividend bcb budget with tax increase xb Income xcSubstitutionxa effect u2 u3 dividend u2 Consumption of other goods, y Consumption of other goods, y u1 mc 375 xd Fossil fuels consumed, x (b) Second income effect due to citizen dividend price of fossil fuels. With the increase in the price from pa to pb due to the the tax, the citizen’s budget constraint becomes steeper. It pivots inward round its y-axis intercept because the amount of the budget itself is unaffected, but the price has increased. So bcb is the new budget constraint. As before, citizen now maximizes her utility now consuming at point b where her marginal rate of substitution equals the new marginal rate of transformation, pb . At b the constrained utility maximum, the citizen has decreased her consumption of fossil fuels from xa to xb , consistent with the policy goals of the tax to decrease consumption of fossil fuels. Setting aside the value that the citizen places on the overall mitigation of the the climate change crisis that the population wide effects of the policy accomplished, the policy has lowered her utility because she is on a lower indifference curve, u1 < u2 . Is it fair? This will be true of all citizens, but the effect will differ across by levels of income. In the U.S., the reduction in real income imposed by the carbon tax on will be a larger among poorer households. This is because, as Figure 7.20 shows: • while (panel a) higher income people spend much more than lower income households on carbon costs (think about air travel, heating and air conditioning large houses) • ...expenditure on carbon costs as a fraction of their total expenditure is greater for lower income households (panel b). As a result high income people will pay more of the tax than low income people, but the tax will lower the real income of poor people by a larger per- Figure 7.19: Carbon tax with dividend. A citizen decides on consumption of fossil fuels (x) other goods (y). Prior to the introduction of the carbon tax the citizen’s budget is ma and the budget constraint is the line labeled bca , with slope pa . The utility-maximizing allocation is on the budget constraint at point a where the citizen’s indifference curve u2 is tangent to the budget constraint. A carbon tax is proposed that increases the price from pa to pb steepening the budget constraint. The result, shown in panel a, is the budget constraint pivots inwards. The effect of the price change – a reduction in fossil fuel consumption from xa to xb – is the sum of the income and substitution effects. In panel b the citizens’ dividend increases the household’s budget so the budget constraint shifts outwards and has a vertical intercept (the budget itself) md > mb . The indifference curves shown here are based on a Cobb-Douglas utility function with a = 0.5. 376 MICROECONOMICS - DRAFT centage that will be the case for higher income people. In the U.S. the carbon tax is regressive, meaning that the amount paid as a fraction of a household’s income is greater for lower income households. This is where the citizen’s dividend comes in. Increasing income and ensuring fairness through a citizen’s dividend To see how the citizens’ dividend alters the result, turn to Figure 7.19 b. As in panel a, the citizen is at point b with lower utility than before the tax. But now suppose the total carbon tax revenues collected are divided equally and distributed equally to each household as in the amount mb to md . This is the citizen’s dividend. As you can see the effect is to shift upwards the budget constraint by the same amount. With the higher income, the citizen maximizes her utility at point d. For the citizen we have modeled, the dividend mb to md is large relative to her previous budget. The result is an increase in consumption of other goods, so that her level of utility is higher than it was before the tax, that is, comparing points d and a, u3 > u2 . At d, they have higher consumption of other goods (yd > yb ) and they consume the a lower level of fossil fuels than they did previously, but greater than before the dividend (xa > xd > xb ). With the greater consumption of other goods, the citizen has higher utility and they obtain utility of u3 > u1 . All citizens will receive the same dividend. This means that more than half of the households – those poorer than the mean income – will experience an increase in their disposable income (income after paying taxes and receiving the dividend) as a result of the carbon tax and citizens’ dividend policy. They will pay less in taxes (which are proportional to the cost of the carbon they consume) than they receive in the citizen’s dividend which is proportional to the mean carbon costs consumed in the population and hence equal for all citizens. While the carbon tax alone is regressive,the carbon tax and the citizen’s dividend taken together is progressive. The case we modeled in Figure 7.19 illustrates the increase in disposable income and utility of a poorer than average citizen. International differences In some countries, however, the picture is reversed, with poor people spending a small fraction of their budget carbon related consumption and wealthy families spending a larger share on fuel as a relative proportion of their income. The data in Figure 7.21 for Mexico show exactly this, at least for motor fuels. As with carbon costs as a whole, the U.S. data (in the left panels) show F AC T C H E C K Using detailed data on household expenditures economists Anders Fremstad and Mark Paul calculated that a $ 50 per ton of CO2 would leave 56 percent of all U.S households with a higher real income (the same would be true of 84 percent of households with incomes less than the mean.)5 D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 377 15% Cost as percent of income Cost in $ per person 4000 3000 Mean 2000 Median 1000 0 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 1 2 Decile (a) Average carbon costs per person by income decile 3 4 5 6 7 8 9 10 Decile (b) Average proportion of expenditure on carbon by income decile that the fraction of the household budget spent on electricity and motor fuels respectively falls dramatically as income rises. But this is not the case for Mexico. The fraction of the household budget spent on electricity is only modestly lower for the upper income households. And for motor fuels (car and truck use) the fraction of the budget spent is much greater for high income households than for those with lower incomes (among whom car and truck ownership is limited). Figure 7.20: Dividend distribution and carbon costs. The left-hand figure shows the absolute amount spent on carbon for each of the ten household income deciles (1 is poorest, 10 is richest). The right-hand figure shows the proportion of household expenditure on carbon by for the ten income deciles. Source: Fremstad and Paul (2019) using data from the Energy Information Agency, the Bureau of Economic Analysis, and the Bureau of Labor Statistics. 7.9 Application: Giffen Goods and The Law of Demand The demand curves you have seen all slope downward: a lower price is associated with more purchases. This is called the Law of Demand, and the movement of prices and quantities purchased in opposite directions that it predicts is widely observed. But there is a special kind of good – called a Giffen good – for which the law of demand is violated. For Giffen goods, a higher price is associated with a greater amount of purchases. You already know that for an inferior good the amount purchased will decline as income rises. This is not really surprising, some of the low cost foods that people eat when they have very limited budgets will not be purchased at all when they have more income to spend. A Giffen good really is surprising because less is purchased when its own price decreases. How could this be? Think about a poor family consuming a large amount of some inferior good. When the price decreases, as you know, there is both a price effect motivating the family to purchase more and an income effect resulting from the decrease in price. Because the good is inferior, the higher real income of the family motivates them to purchase less of the good. If the negative income effect is L AW OF D EMAND The law of demand holds that a decrease in the price of a good will be result in an increase in the quantity of the good purchased. Giffen goods are an exception to the law. 378 MICROECONOMICS - DRAFT Average Electricity Expenditure (% of Total) Average Electricity Expenditure (% of Total) 2.0 6 4 2 0 1.5 1.0 0.5 0.0 1 2 3 4 5 6 7 8 9 10 1 2 3 Expenditure decile 5 6 7 8 9 10 (b) Mexican Consumption of Electricity Average Motor Fuels Expenditure (% of Total) (a) U.S. Consumption of Electricity Average Motor Fuels Expenditure (% of Total) 4 Expenditure decile 6 4 2 0 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 1 2 3 Expenditure decile (c) U.S. Consumption of Motor Fuel 4 5 6 7 8 9 10 Expenditure decile (d) Mexican Consumption of Motor Fuel greater than the positive substitution effect, purchases will decline in response to a decline in the price. Here is an example. In China, for very poor households, rice is the main staple food and if they have enough money, they add other foods that make rice taste better, such as shrimp or beef. However, when the price of rice increases, this means the households have little money left over to buy beef or shrimp. Consequently, they will consume more rice even though it’s price has increased. As a result, over some range of prices the demand curve for rice for these families is upward-sloping. Of course if the price of rice rose so high that the household purchased only rice, then further price increases would have to reduce the amount purchased, so the demand curve would then be downward sloping as the Law of Demand requires. Just such a demand curve Figure 7.21: U.S. and Mexican consumption of motor fuel. Each county’s income distribution is divided into deciles from poorest (1) to richest (10). The average consumption of motor fuel as a proportion of total family income for each decile is shown by the size of the bar for that decile. In the U.S., the consumption as a share of income is higher for lower deciles than for higher deciles. In Mexico, the consumption as a share of income is lower for lower deciles than for higher deciles. Source: Pizer and Sexton (2019) using data from the U.S. Expenditure REM I N DConsumer E R An inferior good Survey is one (Bureau for which of Labor Statistics) Mexico’s National Survey purchases fall asand income rises. A Giffen of Income and Expenditure (Instituto Nacional de good must be inferior, but not all inferior Estadistica y Geografia). goods are Giffen goods. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S is illustrated in Figure 7.22. This is exactly what economists who studied subsidies of rice observed in Hunan, a region of China. They ran an experiment by subsidizing the price of rice, lowering the price the families actually paid. When they provided the subsidy, very poor households reduced their consumption of rice. That is, 379 G IFFEN GOODOver some range of prices, purchases of a Giffen good increase if the price rises, and fall if the price falls. Giffen goods are an exception to the law of demand. p the rice subsidy so that prices rose, the households consumed more rice. For these households rice was a Giffen good.6 7.10 Market demand and price elasticity at that price of all the people making up the demand side of the market. We can compute the market demand by adding up the individual demand curves. If we plot all the demand curves with the quantity demanded of x on the horizontal axis and the price p on the vertical axis, this requires the horizontal summation of the individual demand curves. We use an upper-case X for market demand and a lower-case x for an individual demand. Figure 7.23 shows how (on the left) an individual market demand curve is (on the right) summed over ten people to produce the market demand curve, that is X ( p) = x1 ( p) + x2 ( p) . . . + x10 ( p). A Linear Market Demand Curve (Quadratic quasi-linear utility) If there are n identical buyers, each of whom has the same quadratic, quasilinear utility for the good, with the same parameters x̄, p̄, each individual has the demand (from Equation 7.32): x(m, p) = x̄ x̄ p p̄ (7.39) The market demand is then the sum of all the individual demands. But, since they are all equal for the identical people, this is the same as the the number of people (n) multiplied by the individual demand curves. In the quadratic quasi-linear case, therefore, the market demand curve is a given by the following: Market demand X ( p, m, n) = Number of people ⇥ Individual demands ✓ ◆ x̄ = n x̄ p p̄ X̄ = X̄ p (7.40) p̄ Because the market demand is the summation of individual demands, the market demand function is also downward-sloping: quantity demanded falls as the market price increases. Giffen demand Downward− sloping demand Upward− sloping demand The market demand for a good at any given price is the sum of the demands Individual demand Price per unit, $, p when rice was cheap, households consumed less of it. When they removed x Quantity of the good, x Figure 7.22: Demand for a Giffen good. The demand for a Giffen good is upward-sloping when the price is low. Over this region, a higher price is associated with more purchases. At a sufficiently high price, though, demand becomes downward-sloping. M ARKET DEMAND CURVE The market demand curve is the horizontal summation of individual buyers’ demand curves. That is, for each price (on the vertical axis) we add together each person’s quantity demanded at that price 380 MICROECONOMICS - DRAFT p = 10 p = 20 Price per kilogram, p Price per kilogram, p p = 20 Market price, p = 10 i ● 1 Demand: x(p) = 10 − p 2 Inverse Demand: p(x) = 20 − 2x 5 10 Demand, X(p) = 100 − 5p 1 Inverse demand, p(X) = 20 − X 5 p = 10 ● (b) Market demand Re-arranging Equation 7.40, we can find the inverse market demand function: p(X ) = p̄ p̄ X X̄ (7.41) The inverse market demand curve is linear with a vertical intercept of p̄ (the maximum willingness to pay of buyers like Harriet), a horizontal intercept of X̄ , Dp p̄ . X̄ M-Note 7.9: Market demand with 10 buyers Let us assume that the fish market is made up of Harriet and 9 other buyers who are identical to her (a total of ten buyers). Harriet’s quadratic, quasi-linear demand function was: Harriet’s Demand: x( p) = 10 1 p 2 (7.42) If all the fish buyers are identical to Harriet, then we can sum their demand functions (quantity as a function of price), xi ( p) to get the market demand, X ( p). This is the same as multiplying the demand function by the number of people, n = 10, to get the market demand function: X ( p) = = = = X = nx = 100 Kilograms of fish, X (a) Individual demand and a slope of Dx = Market price, p = 10 X* = 50 Kilograms of fish, x Inverse market demand i n(xi ( p)) 1 n(10 p) 2 1 10 ⇥ (10 p) 2 100 5p (7.43) Recall, though, that we typically graph price as a function of quantity, or the inverse demand function. We use the market demand curve to find the inverse market demand curve with price as a function of quantity by re-arranging Equation 7.43 and similarly for Harriet Figure 7.23: Individual and market demand. In figure a., we present Harriet’s demand at different prices per kilogram of fish. On the right, is the market demand for fish, which is the sum of ten identical fish-buyers’ demands for fish (including Harriet). Notice that Harriet’s individual demand curve is much steeper than the market demand curve. The change occurs because, for example, for every $2 decrease in the price Harriet will buy one more unit of fish, for the market as a whole, each of 10 people would buy one more unit of the good. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 1 p(X) = 20 − X 5 p = 20 Price per unit of the good, p p Price per unit of the good, p Elastic η >1 Unit Elastic η =1 ● Inelastic η <1 pf = 14 ● ⎛6⎞ ηg = 5⎜ ⎟ = 0.43 ⎝70⎠ e pg = 6 ● x Quantity of the good, X ⎛10⎞ ηe = 5⎜ ⎟ = 1 ⎝50⎠ ● Xf = 30 0 ⎛14⎞ ηf = 5⎜ ⎟ = 2.33 ⎝30⎠ f pe = 10 381 g Xg = 70 Xe = 50 X = 100 Market quantity of the good, X (a) General values of price elasticity (b) Price elasticity for 10 People like Harriet in a market with re-arranging Equation 7.42: Harriet’s inverse demand: p(x) = 20 Market inverse demand: p(X ) = 20 2x 1 X 5 (7.44) (7.45) Contrasting Equations 7.44 and 7.45 we can see that they have identical vertical intercepts equal to p̄, but the slopes of the two functions differ. X To see why, notice that x = n , and substituting this expression for x into Harriet’s in inverse demand (Equation 7.44) we get the equation for Market inverse demand (Equation7.45. This is why the market inverse demand curve has a slope equal to the slope of Harriet’s inverse demand namely 2, divided by the number of total buyers (10), for a slope of 15 for the market inverse demand curve. The market demand curve is therefore flatter than Harriet’s relatively steep demand curve. Price elasticity and the slope of the demand curve Figure 7.24: Price elasticity of demand: general and specific cases. In figure a. on the left, we present the general relationship between the demand curve and the value of price elasticity of demand, h . The figure shows how price elasticity varies from a high value to a low value as you move left to right along the demand curve. In figure b. on the right, we present the Market demand curve for 10 buyers like Harriet whose preferences are the horizontal sum of Harriet’s resulting in a market demand curve of p(X ) = 20 15 X . Consequently, we can calculate three values for price elasticity of p demand using the formula of h = dd Xp X . The slope of the curve, Dp dX = 1 5, therefore inverting that value we see that DX d P = 5. We can substitute in the values for p and X at each of the price quantity combination to find the value of price elasticity at each of the points e, f, and g as shown in the figure. For many questions of both firm strategy and public policy an important question is: how much change there is in the quantity demanded when there is a change in price or DX Dp . But this is expressed in two different units: the units of the good (kilos of fish) and the monetary unit (dollars). But we often need to compare responsiveness across commodities – is the demand for restaurant meals more or less responsive than the demand for motor vehicle fuel? To allow for comparisons across commodities, we need a measure of responsiveness to price that does not depend on the units in which it is measured, whether the quantity demanded is in kilos of fish or liters of Coca Cola, whether the price is in Yen or Euros. We therefore describe the response of market demand to a change in price as the ratio of the percentage change /X in quantity demanded to the percentage change in price, DX Dp/p . This ratio is called the price elasticity of demand and often represented by the Greek letter h (pronounced "ai-ta"). P RICE ELASTICITY OF MARKET DEMAND The price elasticity of market demand with respect to price at a point (X, p) is the ratio of the percentage change in quantity demanded to the percentage change in /X price,hX p = DX Dp/p . 382 MICROECONOMICS - DRAFT hX p Price elasticity of demand = = hX p = % Change in price % Change in quantity DX /X Dp/p DX p Dp X (7.46) The price elasticity of demand at any point on the demand curve is equal to the slope of the demand curve multiplied by the ratio of price to quantity at that point. The price elasticity of demand falls into three categories: |h| > 1 Demand is price-elastic, which means that the quantity demanded responds more than proportionally to a change in price. |h| = 1 Demand is unit price-elastic, which means that the quantity demanded responds exactly proportionally to a change in price. |h| < 1 Demand is price-inelastic, which means that the quantity demanded responds less than proportionally to a change in price. Slope and price elasticity with a linear demand curve (the QQL case) As you already know the slope of the quadratic, quasi-linear demand curve is constant. But its price elasticity changes as price and quantity change along the demand curve: h= X̄ p p̄ X (7.47) p The term X̄p̄ is constant, but the term X is very large when p is close to p̄ and X is close to zero, and goes to zero when p is close to 0 and X is close to nx̄ or X̄ . It is tempting to think of the price elasticity of demand for a partic- ular good as a single constant number, but in general the price elasticity of demand changes with price and quantity demanded. M-Note 7.10: Price elasticity along a Linear Demand Curve When there are 10 people with demand functions as in M-Note 7.9, then we can evaluate the elasticity of demand as follows. Remember the following parameters: p̄ = 20; x̄ = 10; n = 10, therefore X̄ = nx̄ = 100. We now evaluate price elasticity of demand at two different (X, p) points. Using Equation 7.47 when price is 14 and quantity demanded is 30 units (refer to Figure 7.24): h= DX p = Dp X X̄ p = p̄ X 5 ✓ 14 30 ◆ = 2.33 Therefore, we would say that for 10 people in the fish market, the price-elasticity of demand is elastic because |h| > 1. Now consider an alternative point with a lower market price, p = 6, and corresponding higher quantity demanded equals X ( p) = 100 5 ⇥ 6 = 70. We then calculate elasticity M - C H E C K The slope of the demand curve is negative, so the elasticity is also a negative number. We usually refer to the absolute value of the elasticity |h|; so a "larger" elasticity means at any given price and quantity a flatter demand curve. 8 D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 6 4 5 ● 2 3 Price per unit of x, p 7 Rice in Japan, η = 0.2 Fish, η = 1.5 Coca−cola, η = 3.8 Expensive alcoholic drinks, η = 4.7 0 1 2 3 Quantity of the good, X as follows: h= DX p = Dp X X̄ p = p̄ X 5 ✓ 6 70 ◆ = 0.43 At the lower price, p = 6 (which is on the lower portion of the demand curve), the priceelasticity of demand is inelastic because |h| < 1, which means that quantity demanded responds less than proportionately to a change in prices. Checkpoint 7.7: The price elasticity of demand a. Return to Figure 7.3 and identify the products (and their price ranges) that are most price elastic and most price inelastic. b. Identify points on the demand curve in Figure 7.24 for which an increase in the price will a) raise, b) lower or c) leave unchanged total revenue. c. Using the Cobb-Douglas demand curve we found in Equation 7.12 find the price elasticity of the good, x, in general (find DX DP to substitute into Equation 8.33). d. Assume that a = 0.5 and m = 100 as in the figures drawn in the examples and then consider the three prices used in Figure 7.9 to calculate the price elasticity of demand at each bundle. 383 Figure 7.25: Comparison of elasticities for different goods. The demand functions shown are called iso-elastic, meaning that unlike the case for linear demand functions, the price elasticity of demand is the same at every point on the curve. (Recall that iso means equal). The functions have the form x( p) = kph . What the figure shows is that, for example, if the quantity of fish demanded when the price per kilo is 5 is 1 kilo, then if the price fell to 3 per kilo the demand would approximately double (increase from 1 to 2). Sources: dhar2005cocacola; Akino and Hayami (1975), Chomo and Ferrantino (2000), and Miravete, Seim, and Thurk (2020). 384 MICROECONOMICS 7.11 - DRAFT Application. Empirical estimates of the effect of price on demand. Why are some goods more price elastic than others? Because we can observe price changes and how the quantity purchased changes as a response, we can estimate the price elasticity of demand for various goods. Some estimates are illustrated in Figure 7.25. The demand for a good will be highly inelastic if it is "something that you cannot do without" and it also does not constitute a large fraction of your budget. F AC T C H E C K The demand for another sugary drink - Mountain Dew – is even more price elastic that Coca-cola, namely, |h| = 4.39. Coke and Mountain Dew are close substitutes. The price elasticity of demand for sugary drinks as a whole is estimated to be much lower, i.e. |h| = 1.4. This is because other drinks, like milk or tea are not really close substitutes for sugary drinks. As a result a price increase for Coke might get you to switch to Mountain Dew, but not to tea or milk. Generalizing from this intuition we expect goods to be price inelastic if: • there are few substitutes for the good in question (e.g. brand loyalty, addiction, or prescription medication) • it is considered to be a necessity not a luxury (e.g. rice, not expensive liquor or Coca-cola) • it is not a large fraction of your total expenditures (e.g. fish) • the person making the decision to buy is not the person paying for the purchase (e.g. when a doctor prescribes a drug that the patient will pay for). San Francisco 0.52 New York City 0.61 Los Angeles Figure 7.26 presents a set of different estimates, for the a single product, but in different U.S. cities. Checkpoint 7.8: Uber elasticities 1. Why do you think the estimated price elasticity of demand for Uber rides in Chicago and New York City are more than twice the estimates for LA. Look up the density (population per square mile or kilometer) for the four cities. Does this provide any clues about why the elasticities might differ as they do? How could evidence about the system of public transportation provide additional clues? The price elasticity of demand for sugary drinks and the effect of a tax Obesity and its associated illnesses inflict extraordinary suffering and mounting health care costs around the world. Among high income countries, the U.S. and the UK are especially hard hit while obesity is rare in Japan and Singapore. Among middle income countries Mexico has one of the highest obesity rates. Among the contributors to the epidemic rise in obesity rates in recent years, economists have proposed, are two facts: 0.33 Chicago 0.66 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Price elasticity of demand Figure 7.26: [ 0cm] Estimates of the price elasticity of demand for Uber rides in four US cities in 2015. The estimates are based on almost 50 million observations on Uber rides from the first 24 weeks of 2015 in Uber’s four biggest U.S. markets. Source: Cohen et al. (2016). D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 385 • As economies shift from farming and manufacturing to services the amount of calories we use in a day’s work has declined and • the cost of calories, relative to other things we might spend our money on, has fallen. Governments around the world have addressed the second economic proposed cause of obesity– reduced cost of calories – by instituting so-called “fat taxes” either to tax the consumption of saturated fats or to tax the consumption of sugar. As of 2019, 7 U.S. cities and 34 countries had implemented such policies. These taxes do not aim to increase government revenue. Instead, the government wishes to discourage citizens from consuming the goods because of concerns over the citizenry’s health. Similar reasoning applies to "sin taxes", which are taxes on cigarettes and liquor to discourage excessive consumption of those goods. Sugary drinks that are commonly taxed in many countries include: • fruit drinks (which includes sports drinks and energy drinks), • pre-made coffee and tea (for example, bottled iced coffee and iced tea), • carbonated soft drinks, and non-carbonated soft drinks (which includes cocktail mixes, breakfast drinks, ice pops, and powdered soft drinks) The average American household consumed an average of 156 liters of these drinks per year during the years 2007-2016. The demand curve and the price elasticity of demand derived from it provide essential pieces of information to assess the likely effects of the tax on sugary drinks. Because retailers frequently change prices of their drinks it is possible to estimate the price elasticity of demand. To do this, a team of economists recorded sugary drink sales, prices, and a long list of other possible influences on the individual’s purchases (including health information, how much they "liked" sweet drinks, and more). On this basis they estimated that the price elasticity of demand for sugary drinks is about - 1.4, meaning that a ten percent increase in the price of the drinks would result in a 14 percent decrease in demand. This estimate – for sugary drinks as a whole – is much less than for one particular drink (Coca Cola) because there are many other sugary drinks that are close substitutes for Coca Cola. Figure 7.27 illustrates what an elasticity of this magnitude implies. In this example we do not ask what determines the price per liter p0 . Instead we ask the hypothetical question: how many liters would be demanded at various prices? We can study the effect of a sugary drinks tax on the amount con- F AC T C H E C K The fact that obesity rates (BMI > 30) in the U.S. are ten times the rate in Japan suggests that these economic factors – which apply with about equal force in the two countries – cannot be the entire story. Differences in culture and public policies also matter.7 386 MICROECONOMICS - DRAFT Price per liter, p Figure 7.27: The effect of a price increase on the demand for sugary drinks when the price elasticity of demand is |h| = 1.4. The demand curve shown is iso-elastic with a price elasticity of demand of |h| = 1.4. There are two prices, the pretax price ( p0 ) and the post-tax price ( p1 = p0 + Dp ) where Dp = 0.20xp0 . At the higher price, the consumption of sugary drinks is less, comparing x0 at point a to x1 at point b. p1 = p0 + Δpτ = $1.50 ● b p0 = $1.25 ● a Demand x1 = 108 x0 = 150 Quantity of sugary drinks (liters), x sumed by comparing the price per liter without the tax to amount consumed at the the (higher) price when the tax is imposed. Suppose the price of sugary drinks was initially $1.25 dollars per liter. At this price we can see that the typical person would have demanded about 150 liters. Figure 7.27 depicts this interaction. The starting price and quantity combination are shown by (x0 , p0 ). If the effect of the tax was to raise the price of sugary drinks by 20%, the the price after the imposition of the tax would be 1.50 per liter. Recall that with a price elasticity of h = 1.4, this means a 10% increase in the price will result in a 14% decrease in quantity demanded. So, the effect of a 20% increase in the price is that the quantity demanded will decrease by 28% , down to 108 liters ( x1 in the figure) after the tax. Checkpoint 7.9: Why is demand for sugary drinks price elastic? Explain why the demand for sugary drinks as a whole in the U.S. is less price elastic than the demand for Coca Cola or Mountain Dew. 7.12 Consumer surplus and interpersonal comparisons of utility When a person, call her Harriet as we did earlier, buys a good she does so because she expects to derive a benefit that exceeds the price of the good. The difference between the most she would be willing to pay for the good and C ONSUMER S URPLUS is the difference between a person’s willingness-to-pay and what they actually pay for each unit of the good that they consume. Because not purchasing the good is the buyer’s fallback option, this definition shows that consumer surplus is a rent. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 20 18 Price per unit ($), p 16 14 Consumer surplus 12 Market price, p = 10 p = 10 8 6 4 Buyer's demand or willingness to pay 2 0 0 1 2 3 4 x=5 6 7 8 9 10 Quantity, x what she actually pays for it is called the consumer surplus that she received as a result of that purchase. Because it is measured as the difference between the maximum willingness to pay in money and the money that is paid, consumer surplus gives a measure in monetary terms of the benefits (or “ consumer welfare”) that a person derives from a purchase of a good. If we consider not buying the good as the person’s fallback position, then we see that consumer surplus is an economic rent, namely a measure of what they get above and beyond her next best alternative. If we assume that the marginal utility of a unit of expenditure – the value to the buyer of having one more Euro to spend, for example – is the same across all people, then consumer surplus can be summed over all the people who buy the good. Total consumer surplus summed over all buyers of the good measures the increase in well being to people that is made possible by the availability of the good. Figure 7.28 illustrates the consumer surplus available to Harriet by her purchases of 5 units of good x. The maximum she would pay for the first unit – think: access to a workout at the gym during a week, a film on Hulu – is $20. But each successive unit is worth less to her, and if she already has 4 – workouts, films – the most she would pay for the 5th is $12. The sixth would not be worth more than she’d pay for it. So she will purchase 5 units. Adding up her willingness to pay for each and subtracting what she actually paid – $10 in each case she has a consumer surplus of $30 (that’s 10 + 8 + 6 + 4 + 2 = 30). 387 Figure 7.28: Harriet’s willingness to pay and consumer surplus. The height of the steps in the step function is the maximum Harriet would pay to have the first, second, third, and so on unit of the good. For each unit she buys, she pays $10. Her consumer surplus is the vertical distance between the price line at p = 10 and her willingness to pay for each additional unit, summed over the number of units she purchases. Her utility-maximizing consumption bundle is x = 5, when her willingness to pay equals the price: mrs(x) = p = 10. Summing over the bars, she receives consumer surplus, CS = 10 + 8 + 6 + 4 + 2 = 30. 388 MICROECONOMICS - DRAFT The graph of her willingness to pay – called a step function – is her demand curve. When we think of people’s purchase over a longer period of time, or the purchases of many people, we smooth out the ’steps’ and make a smooth curve (not necessarily a straight line). We are often interested in a measure of how much people as a group benefit from the opportunity they have to purchase some good. A natural way to do this is to add up the consumer surplus enjoyed by each buyer. For this to make sense it must be that a dollar’s worth of consumer surplus is as valuable to one person as to another. Unless we assumed this, we could not add the consumer surplus of one person – some dollar amount – to some other person’s consumer surplus. There are really two parts of this key assumption: • We can make inter-personal comparisons of utility : we can compare one person’s well being (or utility) with another. Recall that this means that we consider utility to be a cardinal measure that can be compared across individuals (like for height, how much taller is Simon than Harriet) rather than an ordinal measure (Simon is taller than Harriet). • The marginal utility of money left over for other purchases is the same to all people: an additional dollar makes the same contribution to the well-being of one person as to another. This means that what Harriet would purchase with a dollar of money left over after spending on workouts contributes as much to her well being as a dollar’s worth of additional expenditure by a less fortunate person. This would almost certainly not be true if one of them were very poor, so that a dollar would be worth a lot (it would be used to purchase food, or other essentials), and the other was very rich (the additional dollar would be spent on a luxury good). In the left panel of Figure 7.29, the individual consumer surplus (Equation 7.48) is the area of the light-yellow triangle above the price and below the demand (marginal rate of substitution) function for Harriet. We can see that consuming x = 5 units of fish provides Harriet with consumer surplus of $25. Notice that this is exactly the same as if we had substituted Harriet’s values for x̄, p̄, p and x into equation 7.48. The consumer surplus for all buyers is the area shaded in yellow in the right panel of the figure and because we have assumed all buyers are identical this is exactly n = 10 multiplied by the individual consumer surplus for Harriet. Notice that the scale of the x-axis is ten times larger, which is consistent with Equation 7.49. E X A M P L E To measure the utility gained by making a purchase and to sum this across individuals we need a measure of utility that is similar to money. If one person has a thousand U.S. dollars and someone else a hundred U.S. dollars we can say that the first person has ten times as much money as the second, irrespective of whether we measure their wealth in dollars, or in pennies. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S p = 20 Consumer surplus p = 10 Market price, p = 10 i Consumer expenditure Price per kilogram, p Price per kilogram, p p = 20 Consumer surplus m p = 10 Market price, p = 10 Consumer expenditure Inverse demand 1 p(X) = 20 − X 5 Inverse demand p(x) = 20 − 2x xi x Xm = 50 Quantity of fish in kilograms, x (b) Market consumer surplus M-Note 7.11: Consumer surplus with quadratic, quasi-linear utility The demand curve based on quadratic, quasi-linear preferences is linear. Therefore, we can use basic algebra and geometry to calculate the value of the consumer surplus. We use the following three data points: • The person’s maximum willingness to pay is p̄. • The person actually pays the price p < p̄. • Therefore the difference between what they pay and what they are willing to pay is p̄ p for the first unit consumed. • For units after the first unit consumed, the willingness to pay decreases, and the consumer surplus equals the difference between their willingness to pay at that unit and the price they actually pay, i.e. CS(x) = mrs(x) p. • They consume xi at the price, p. • Therefore the consumer surplus of what they consume is the area of the triangle between p̄, p and xi , which has an area of 12 ( p̄ p)xi = csxi Therefore, consumer surplus for the individual and the market is given by: Individual Market cs(x) = CS(X ) = = nx = 100 Market quantity of fish in kilograms, X (a) Individual consumer surplus 7.13 389 1 ( p̄ p)x 2 ✓ ◆ 1 ( p̄ p)x ⇥ n 2 1 ( p̄ p)X 2 (7.48) (7.49) Application: The effect of a sugar tax on consumer surplus Some taxes like the so-called sin taxes on cigarettes and liquor – do not aim primarily to raise government revenues – the usual motive for taxation. Instead sin taxes aim to alter people’s behavior: to discourage smoking, drinking Figure 7.29: Individual and market consumer surplus. On the left, we present a version of figure ?? showing the utility-maximizing kilograms of fish Harriet buys and the consumer surplus she derives as a consequence. She consumes output, x, and therefore her cs(x) = 12 ( p̄ p)x = $25. Her expenditure is the price she paid multiplied by the number of units she bought: ce(x) = p(x) ⇥ xi . On the right, is the market demand for fish with the market consumer surplus which is the net benefits that all people obtain from paying prices beneath their maximum willingness to pay, thus cs(X ) = 12 ( p̄ p)Xm = $250. MICROECONOMICS - DRAFT Pre−tax consumer surplus a p0 = $1.25 Pre−tax expenditure Demand Price per liter, p Price per liter, p 390 A p1 = p0 + Δpτ = $1.50 p0 = $1.25 x0 = 150 Quantity of sugary drinks (liters), x (a) Initial consumer surplus and expenditure alcoholic beverages, and consuming foods that contribute to obesity. We showed earlier in Figure 7.27 when discussing price elasticity of demand that an increase in prices through a tax will decrease quantity demanded by people. We now analyze the consequences of that tax for people’s utility. Figure 7.30 shows the consequences of the tax for consumer welfare (measured in terms of prices and quantities consumed). Figure 7.30 a. shows how, at the initial price, consumer surplus is given by the combined area in green, shown by the total area A + C + E in Figure 7.30 b. After the tax at the new higher price, consumers will be left with consumer surplus equal to area A. Consumers will lose consumer surplus indicated by the area E + C. Area B is the portion of consumer expenditure that is unchanged by the tax. Consumer expenditure will decrease by the area D, but increase by the area E. That is, before the tax, consumers spent areas B + D. After the tax, consumers spend areas B + E. In Chapter ?? we will return to the sugary drinks tax, looking at its impact on on others, including firms’ owners who will lose economic profits as a result. Is it fair? Sugary drink taxes are regressive In 2017 voters in Santa Fe, the capitol of the U.S. state of New Mexico, voted overwhelmingly to reject a proposed tax on sugary drinks. The measure had been put forward by a popular mayor and would have directed the resulting revenue towards expanding pre-school educational opportunities for the less well off. It was opposed by the American Beverage Association. Opponents of the measure held that unfairly placed a burden on the less well b E C B D a Demand x1 = 108x0 = 150 Quantity of sugary drinks (liters), x (b) Consequences of the tax Figure 7.30: Effects of a tax on consumer surplus and expenditure. There are two prices, the pre-tax price ( p0 ) and the post-tax price ( p1 ). As the price increases, the consumption of sugary drinks decreases from x0 at point a to x1 at point b. In Figure a. the area shaded in green below the demand curve and above the price, p0 , is the pre-tax consumer surplus. The area shaded in blue is the pre-tax expenditure, equal to the price multiplied by the quantity people consumed, i.e. p0 x0 . Area A is the consumer surplus that remains after the imposition of the tax. Area B is the expenditure that was common before and after the imposition of the tax. Area D is the decrease in expenditure by people who are unwilling to purchase sugary drinks at the higher price, p1 . Area E was consumer surplus before the tax, but is now part of expenditure. Area C plus area E is the decrease in consumer surplus as a consequence of the tax. Liters purchased per adult equivalent per year D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 391 Figure 7.31: Consumption of sugary drinks in households of differing incomes. The figure shows the amount of sugary drinks purchased per ‘adult equivalent’ per year by income (measured in thousands of dollars). The term ‘adult equivalent’ means that children in the households have been counted as some fraction of an adult. The data indicate that the quantity of sugary drinks consumed in the poorest households is about double that consumed in the richest. Source: Allcott, Lockwood, and Taubinsky (2019). 100 90 80 70 60 50 6 16 26 35 46 56 65 85 125 Household Income ($000s) off. To address the potential unfairness of the tax the Santa FE advocates of the sugar tax had linked the measure to the provision of a particular public service that was very much in demand among lower income Santa Feans. But the very real substantial negative income effect apparently outweighed the promise of better educational opportunities. Source: Reporting in The Albuquerque Journal„ the Santa Fe New Mexican, and The Santa Fe Reporter (e.g. "Sugar Tax Fails" 2 May, 2017. Figure 7.31 provides evidence about the consumption of sugary drinks in households of differing incomes based on matched data on purchases of sugary drinks and household income. It is clear that households with lower incomes consume larger quantities of sugary drinks than households with higher incomes do. As a result, a per-unit tax on sugary drinks is regressive poorer households will pay more as a share of their household income. At the same time, however, decreasing the quantity of sugary drinks consumed by members of those households could be quite beneficial for health and for medical costs that those household incur. We will return to the analysis of a sugary drinks tax in the next chapter, taking account not only of the consumer surpluses lost but also the profits lost by owners of sugary drinks producing firms and the benefits made possible by the government revenues raised. Checkpoint 7.10: Policies to mitigate the income losses of less well off people imposed by the regressive sugary drink tax The citizens’ dividend – returning the tax revenues collected to citizens as an equal lump sum payment to each family – is proposed as a way to counteract 392 MICROECONOMICS - DRAFT the regressive nature of the carbon tax. Explain why something similar would not accomplish this purpose in the case of the sugary drinks tax. Experiences around the world of sugar and fat taxes Denmark instituted a per-kilogram tax on saturated fats in 2011. Hungary introduced both sugar and fat taxes in 2011, where the percentage of the tax is proportional to the amount of sugar or fat in the good. In 2012 France introduced a tax on both added-sugar and artificially sweetened drinks of e 0.075 per liter (in 2015). Chile adopted a tax in 2015. In the United States, several states and cities have implemented soda taxes, such as the tax implemented in San Francisco, CA in 2014. The aim of the tax is not primarily to obtain tax revenues but to reduce consumption of the offending foods, so as to improve individuals’ health and to reduce the cost burden of healthcare provision, including by the government. What has happened as a consequence of these taxes? Did the taxes achieve the governments’ aims? Preliminary results from a study in Mexico showed that a 6 Peso-per-liter tax on beverages with added sugar reduced the quantity demanded by between 6% and 12% over the year of the study (2014). Consumption decreased more among low income families with the proportional decrease being between 9% and 17% over the year. The evidence also suggests that consumers switched to close – un-taxed – substitutes that did not contain added sugar such as diet sodas, 100% fruit juices, and sparkling and plain water (with between 7% and 13% increases in these categories).8 In Hungary, the tax has had several effects. The tax has reduced consumption, it has also caused firms to change the recipes of their food items, a sensible response because the tax is proportional to the amount of the sugar or saturated fat the food item contains.9 Denmark’s case is more complicated. The "fat tax" definitely reduced consumption of butter, margarine and similar products, by 10 to 15%. People also changed their buying habits in terms of where they bought their butter and margarine: they switched to buying at discount stores. But, because these stores were aware of these buyer responses and the resulting positive shift in the demand curves they faced, they increased their prices on butter and fatty products more than high-end supermarkets did.10 The tax was unpopular in Denmark and was eventually repealed. Why? People had been crossing the nearby Swedish and German borders to do their shopping: one study showed almost half of Danish shoppers had gone across a border to avoid the tax. These results illustrate the complexity of tax policy when the goal is to reduce D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S consumption of a good. But, in places like Mexico, Denmark and Hungary, we’ve seen significant and important decreases in the consumption of sugary drinks and fatty foods. In Mexico, particularly, this is important for many poor people who are disproportionately affected by health problems caused by high sugar consumption, especially when they cannot afford proper treatment of cardio-vascular diseases or obesity. The experience with taxes designed to motivate healthier consumption shows that these are one tool, among many, in the economist’s toolboox to help curb unhealthy consumption while providing additional funds for public education about diet. Checkpoint 7.11: Salt taxes & Sin Taxes: Putting the elasticity of substitution to work Centuries ago in China, France and the British colony of India the salt tax was one of the major sources of government revenue. What is it about salt that made this tax a good way of raising revenue? Explain why sin taxes levied on goods with price elastic demand (alcoholic spirits, for example) will be effective in changing peoples behavior, but not in raising revenue, while the opposite is true for goods with inelastic demand (for example cigarettes). 7.14 Application. Willingness to pay (for an integrated neighborhood) In Chapter 1 we illustrated the idea of a Nash equilibrium and the process by which a group of people might arrive at such an outcome by the buying and selling of homes among "Blues" and "Greens." We showed that: • The equilibrium composition of the neighborhood – one in which none of the residents wished to switch their location and were able to do so – could be complete "segregation" of the Blues and Greens, even though everyone preferred an integrated outcome. • Which of the multiple equilibria that would be realized was path dependent, like whether the farmers in Palanpur planted early or late: which equilibrium occurred depended on the recent history of the neighborhood. These two characteristics – Pareto inefficiency and path dependence – will be results in the model we now introduce. But here we explicitly introduce a market in homes and people’s willingness to pay. In Milwaukee, Los Angeles, and Cincinnati towards the end of the last century over half of white residents, when asked, said they would prefer to live in a neighborhood in which 20 percent or more of their co-residents were African-American (one in five preferring equal numbers of each).11 But few residents of these cities lived in integrated neighborhoods. Their preferences 393 394 MICROECONOMICS - DRAFT were elicited as part of court records in litigation concerning housing segregation in these and other cities. Most African Americans preferred fifty-fifty neighborhoods. There are many reasons why members of a society might not want their residential communities to be highly segregated. Segregated living leads to racially segregated schools, friendships and other social networks. Because group members would then be unlikely to have friends in the other group, segregated living could encourage group stereotypes and intolerance leading to conflicts between groups. The respondents in the above surveys may have misrepresented their preferences, of course, but those sincerely seeking integrated neighborhoods would have been disappointed. The housing market in these cities produced few mixed white-African American neighborhoods even though these were apparently in substantial demand. In Los Angeles, for example, virtually all whites (more than 90 percent) lived in neighborhoods with fewer than ten percent African American residents, while seventy percent of Blacks lived in neighborhoods with fewer than 20 percent whites. Why was the result at the neighborhood level so seemingly at odds with the distribution of preferences of the individuals making up the neighborhoods? Imagine your surprise had we reported that one in five wanted a back-yard swimming pool and were prepared to pay the price for a pool, yet almost none had pools. Why does willingness to pay get you a pool if you want one, but not an integrated neighborhood? To answer these questions we need an explanation of how highly segregated neighborhoods result, even when preferences are such that members of all groups would be better off with greater integration. In other words we need to understand why the housing market produces a Pareto inefficient level of segregation. Residential segregation is the result of many aspects of how credit and housing markets work, and these differ across countries and even within the U.S among cities and states. But there is another less obvious and perfectly legal way that segregated neighborhoods are sustained, even when most people would prefer a more integrated community. Preferences for integration or segregation We will explain why this is true by modeling a single neighborhood (one of many in a large city) in which, when considered in isolation, all houses are equally desirable to all members of the population (they’re identical). Peoples’ preferences for living in this neighborhood depend solely on the racial composition of the neighborhood. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 395 As before “greens” and “blues” are two population groups that are equally numerous in the city. Greens prefer to live in a mixed neighborhood with slightly more greens than blues and blues correspondingly do not prefer segregation, but prefer a neighborhood with somewhat more blues than greens. We have normalized the size of the neighborhood, setting it equal to 1, so we can refer either to the fraction of greens or the number of greens by g (for fraction), which can vary from 0 to 1. with for example g = 0.3 meaning that the neighborhood is 30 percent green and 70 percent blue. We will express the preferences of the greens and the blues by the maximum prices pG greens and pB blues would be willing to pay for a house in the neighborhood, each depending on the fraction of homes in the neighborhood occupied by greens g. The following willingness to pay (WtP) equations are a way to express the preferences described above: Blues’ WtP pB ( f ) = Greens’ WtP pG ( f ) = 1 (g + d ) 2 1 (g d ) 2 1 (g + d )2 + p 2 1 (g d )2 + p 2 (7.50) (7.51) where p is a positive constant reflecting the intrinsic value of the identical homes. Figure 7.32 shows the willingness to pay equation for the greens, with a low willingness to pay for a house in an all blue neighborhood ( pG (g = 0)), a greater willingness to pay for a house in an all green neighborhood ( pG (g = 1)), but the greatest willingness to pay in an integrated but green-majority neighborhood (with sixty percent greens). The term d is a measure of the preference for segregation. We assume that greens and blues have similar preferences to live with their own group members, so d is the same for the two groups. To see how d measures the degree of preferences for segregation, think about what would be the ideal neighborhood for a green and for a blue. If the ideal neighborhoods of members of the two groups are very different, then segregationist preferences are strong. Because the willingness to pay for a home – and therefore its value – depends on the composition of the neighborhood, the ideal neighborhood has a group composition that maximizes the value of owning a house in the neighborhood (or what is the same thing, that maximizes willingness to pay for a home there). What would each type of person’s ideal neighborhood look like? (M-Note 7.12 explains how these are derived). • Greens: The ideal neighborhood for greens (that which maximizes pG ) is composed of g = 12 + d per cent greens • Blues: Blues prefer an ideal neighborhood with g = 12 d. As the difference between the ideal neighborhoods (that for which they would pay the highest price of a home) of the greens and the blues is 2d we will H I S TO RY This way of thinking about segregation in residential neighborhoods was developed by Thomas Schelling, a Nobel Laureate in economics. You can run a computer simulation of how a population may segregate itself, even with very modest preferences for segregation, here ncase.me/polygons/. 396 Green's willingness to pay, pG MICROECONOMICS - DRAFT increasing willingness to pay as g → 0.6 Figure 7.32: An illustration of the Willingness to Pay of Greens, pG . Their willingness to pay reaches a maximum at point h where the proportion of greens gh = 0.6. Between g = 0 and g = 0.6, Greens’ willingness to pay is increasing as they move from being a minority to become a slight majority at 60% of the population. Between g = 0.6 and g = 1, Greens’ willingness to pay is decreasing as they move from being a slight majority at 60% of the neighborhood to being 100% of the neighborhood. For this and the next figure we used d = 0.1. decreasing willingness to pay as g → 1 pG h pG pG(g = 1) pG(g = 0) gh = 0.6 0 1 Fraction of Greens in the neighborhood, g refer to d as the preference for segregation of the two types (d could differ between the two groups, or one group might not care about the racial composition at all, of course). The willingness to pay curves and the degree of preferred segregation they express provide the essential building blocks for understanding how the housing market will work. But to put that information to work that we need to turn to how the market will change or not depending on its composition. This means we need to identify the Nash equilibria of the market (where there would be no forces changing the situation) and the points that are not Nash equilibria, in which people could do better by buying or selling a house in a way that changes the composition of the neighborhood. This is called an analysis of market dynamics , that is how markets change. M-Note 7.12: Finding the preferred proportions We would like to find the proportion of greens in the neighborhood that would maximize each type’s willingness to pay. We already can see that in the figure this is 60 percent for the greens. To see how we got this number we differentiate Equations 7.50 and 7.51, with respect to g and set the result equal to zero. This gives the value of g that maximizes the greens and blues respectively willingness to pay for a house in the neighborhood. Greens: pG g ⌘ d pG dg = 1 2 (f d) (7.52) Blues: pBg ⌘ d pB dg = 1 2 (f +d) (7.53) R E M I N D E R We discussed dynamics in Chapter 5 when exploring how the fishermen reached Nash equilibrium by comparing their marginal benefits and marginal costs. D E M A N D : W I L L I N G N E S S TO PAY A N D P R I C E S 397 Now, to find the g that maximizes the house value for the two groups, we set each of Equations 7.52 and 7.53 equal to zero and isolate g: Greens: gG max = Blues: gBmax = 1 +d 2 1 d 2 (7.54) (7.55) As can be seen, on either side of g = 0.5 lie the two types’ preferred proportions of green. They are separated by 2d , that is, twice the degree of preferences for segregation. Checkpoint 7.12: Color-blind and other preferences about segregation With the willingness to pay curve of the blues as shown in the figure, draw a new willingness to pay curve of the greens based on an alternative assumption: greens do not care at all about the composition of the neighborhood and that the fixed value they place on homes there is greater than the value that blues place on an all-blue neighborhood, but less than the value that blues place on their ideal neighborhood. If this were the situation, what color-compositions (meaning values of f ) would you expect to see? 7.15 Application: Market dynamics and segregation Remember, an equilibrium is defined by the absence of change. So to determine what level of integration or segregation we would expect to observe (the equilibrium) we need to better understand the process by which the neighborhood composition will change as a result of the dynamics of the market. Home sales: A pathway to segregation To do this, we now consider the conditions under which a house inhabited by DYNAMICS refers to how some market or other economic entity changes. a green might be sold to a “blue family”, or vice versa. Imagine that prospective buyers from outside the neighborhood visit the neighborhood and just knock on the door of a randomly selected house. A sale takes place as long as the house is worth more to the visitor than it is to its current owner. If the current owner values it much more highly, no sale takes place. So houses never change hands among the same types (because they value the houses identically.) But if a green visits the house of a blue, a sale will take place if pG > pB and not if pG pB . Remember that the homes are identical. While the residents in the neighborhood care about the composition of the neighborhood, they are “color blind” when it comes to buying or selling houses: they sell if they are offered a price above what their home is worth to them, irrespective of the color of the buyer. R E M I N D E R This is exactly how we